■fr o 
 
 
 
 
 0^ .-•.,'^o, /\vw/\ c°^-^,;:>o /^^;^/V /.^'-^'^ ^o .-^-^'v''* 
 
 <r 
 
 
 
 
 
 .^^^. 
 
 
 
 i.V-»«, 
 
 .v>. -.i^l^.- ^♦"«$i 
 
 
 %'^*:^^' y "V^-^*/ '^^^^'^r^'y^ '^^^^•^''^^'^o' %'^^^^^\^'' -o^ *.v, , 
 
 ^^ U .^^ rV\^^/^\ v./ /Ji!^\ \,^^ ;^'o \/ ;^\ V^^ ,*;^^^'*- -'^ ^* 
 
 
 
 
 
 0^ ^^ -T^^^^/v 
 
 '• *-/ •^'- ^^-^ •'^'' "■^■'^^ -''A' --^^ 
 
 
 \ %/ 
 
 
 >^-o;c'-\/ V*'^-*V V"^^'*<^^^ ^^^'^-^•'^♦^ .. •<>. 
 
 
 0^ ° ''^^''. '^o \> 
 
 
 
 
 
 /\ '^^S ^ ^^ '^W.^ /\ -.^P/ ^^'^^^ 
 
 
 ■4 .<L*' **,{. - 
 
 

 
 
 .,-&■" .1 
 
 
 '^bv^ 
 
 
 
 
 
 •^0^ 
 
 
 
 
 C" ^\ 
 
 '^^.'^ 
 
 
 • "^-^^f .m\ %.** .•» \/ :A' %.^' •'*^'- '' 
 
 
 
 '^0^ 
 
 'bv'^ :^ 
 
 
 
 
 
 .0^ "^'^^•':^*;^^ 
 
 
BUREAU OF MINES ^?.^ 
 INFORMATION CIRCULAR/1989 
 
 Framework for Autonomous Navigation 
 of a Continuous IVIining iVIachine: 
 Face Navigation 
 
 By Donna L. Anderson 
 
 UNITED STATES DEPARTMENT OF THE INTERIOR 
 
Information Circular 9214 
 
 Framework for Autonomous Navigation 
 of a Continuous IVIining IVIachine: 
 Face Navigation 
 
 By Donna L. Anderson 
 
 UNITED STATES DEPARTMENT OF THE INTERIOR 
 Manuel J. Lujan, Jr., Secretary 
 
 BUREAU OF MINES 
 T S Ary, Director 
 

 Library of Congress Cataloging in Publication Data: 
 
 Anderson, Donna L. 
 
 Framework for autonomous navigation of a continuous mining machine: face 
 navigation. 
 
 (Bureau of Mines information circular; 9214) 
 
 Includes bibliographical references. 
 
 Supt. of Docs, no.: I 28.27:9214. 
 
 1. Coal-mining machinery. 2. Vehicles, Remotely piloted. I. Title. II. Series: 
 Information circular (United States. Bureau of Mines); 9214. 
 
 ^^295,114 
 
 [TN813] 
 
 622 s [622'.334] 
 
 88-600437 
 
CONTENTS 
 
 Page 
 
 Abstract 1 
 
 Introduction 2 
 
 Information and control requirements 2 
 
 Tasks of autonomous navigation in mines 2 
 
 Navigational tasks of continuous mining machine 2 
 
 Face navigation of continuous mining machine 2 
 
 Ahgnment at face 3 
 
 Guidance during cutting sequence 4 
 
 Local navigation of continuous mining machine 4 
 
 Traunming through entry 4 
 
 Turning corner 4 
 
 Global navigation of continuous mining machine 4 
 
 Generate path 4 
 
 FoUow path 4 
 
 Update map 6 
 
 Previous research and attempts in remote and autonomous navigation of continuous mining machines .... 6 
 
 Laser alignment sensor for continuous mining machines 6 
 
 Assessment of optic guidance and supervisory methods 6 
 
 Preliminary design of automated continuous mining machine with remote override 7 
 
 Conclusions of research 7 
 
 Face navigation 7 
 
 Reference for desired cut 8 
 
 Guidance during cutting sequence 8 
 
 Local navigation 8 
 
 Find path 8 
 
 Follow path 9 
 
 Global navigation 9 
 
 Generate path 9 
 
 Follow path 9 
 
 Summary 9 
 
 Sensory system, data manipulation, and control programming 10 
 
 Introduction 10 
 
 Autonomous face navigation of continuous mining machine 10 
 
 Reference for desired cut 10 
 
 Description of guidance from MCS 11 
 
 Benefits of guidance from MCS 11 
 
 Disadvantages of MCS reference system 12 
 
 Navigational guidance system 12 
 
 Sensory solution 12 
 
 Determining position and heading using laser-based scanning system 13 
 
 Determining lateral position and heading using ultrasonic ranging units 16 
 
 Monitoring system 16 
 
 Integration system 17 
 
 Navigational goal scheduler 17 
 
 Action planner 18 
 
 Navigational test-beds 18 
 
 Conclusions 20 
 
 Appendix.-Research issues in autonomous navigation 21 
 
ILLUSTRATIONS 
 
 Page 
 
 1. Typical room-and-pillar mining section 3 
 
 2. Maneuvering for straight-cut 5 
 
 3. Maneuvering sequence for crosscut 5 
 
 4. Mobile roof support 10 
 
 5. Continuous miner advancing face under MCS 11 
 
 6. Block diagram of the hardware-software systems and information transfer for proposed navigational 
 
 guidance system 13 
 
 7. Method 1 for determining position and heading of continuous miner relative to MCS 14 
 
 8. Mathematical representation of method 1 for position determination 14 
 
 9. Method 2 for determining position and heading of continuous miner relative to MCS 15 
 
 10. Mathematical representation of method 2 for position determination 15 
 
 11. Joy 16CM available continuous mining machine 19 
 
 
 UNIT OF MEASURE ABBREVIATIONS USED IN THIS REPORT 
 
 7h 
 
 degree per hour 
 
 h 
 
 hour 
 
 Vmin degree per minute 
 
 in 
 
 inch 
 
 ft 
 
 foot 
 
 ^ min 
 
 minute 
 
 I 
 
FRAMEWORK FOR AUTONOMOUS NAVIGATION 
 
 OF A CONTINUOUS MINING MACHINE: 
 
 FACE NAVIGATION 
 
 By Donna L. Anderson^ 
 
 ABSTRACT 
 
 The first section of this U.S. Bureau of Mines report includes a discussion of the navigational tasks 
 performed by operators of mobile mining equipment, with the focus on the navigational information and 
 guidance techniques required by continuous mining machine operators. Some previous research and 
 attempts in remote £md autonomous navigation of continuous mining machines are mentioned. Issues 
 of research in autonomous navigation of various other mobile robots are presented, with attention 
 focused on their applicability to a continuous mining machine navigating in the mining environment. 
 Conclusions are stated on methods of automating the navigational tasks of a continuous mining machine, 
 and a decision to concentrate initial attempts on the tasks of autonomous face navigation is defended. 
 
 The second section of this report includes a presentation of the Bureau's proposed solution for 
 autonomous face navigation of a continuous mining machine, which includes the employment of a mobile 
 roof support as a reference for guidance in the face area, and a navigational guidance system for the 
 continuous mining machine. 
 
 ^Electrical engineer, Pittsburgh Research Center, U.S. Bureau of Mines, Pittsburgh, PA. 
 
INTRODUCTION 
 
 This report documents the current status of the Bureau 
 of Mines' research into autonomous navigation in mines. 
 The purpose of this research is to support the Bureau's 
 development of autonomous mining operations, which will 
 result in an increase in the overall safety and productivity 
 of current mining operations. Automation will permit the 
 relocation of the mine worker to less hazardous areas and 
 duties in a mine. 
 
 Autonomous navigation is a significant task in the 
 automation of mining processes owing to the need for 
 constant relocation of mine equipment in various work 
 areas of mines. Automatically controlled mobile mining 
 equipment has the responsibility of navigating safely cind 
 
 accurately throughout the immediate work area and be- 
 tween work areas as if a human operator were present. 
 
 To accomplish safe aind accurate navigation in a mine, 
 the equipment must have a knowledge of the area to be 
 traveled along with the capabiUties of determining a path 
 in that area £ind controlling its movement along the path. 
 A knowledge of the area is available via a model of the 
 mine plan, which can be preprogrammed in memory, and 
 a model of the immediate area, which can be obtained 
 from sensory input. Intelligent software can analyze these 
 models and determine traversable paths in these areas. 
 Additional sensory input can be collected to monitor and 
 control the mine equipment's movement along its path. 
 
 INFORMATION AND CONTROL REQUIREMENTS 
 
 TASKS OF AUTONOMOUS NAVIGATION IN MINES 
 
 The navigational tasks in mines can be classified into 
 three types: face, local, and global. These tasks are 
 determined by observing some of the navigational duties 
 necessary for each type. 
 
 The job of face navigation is the positioning of mobile 
 mining equipment in the working face area. Before the 
 cutting process begins, the continuous mining machine is 
 positioned at the face in the proper alignment. A shuttle 
 car is positioned under the conveyor boom of the continu- 
 ous miner to receive the coal cut from the face. During 
 the execution of the cutting process, the continuous miner 
 is guided through a sequence of operations and positions, 
 which lead it to cut the appropriate area as designated by 
 the mine plan. The shuttle car advances as the continuous 
 miner advances and remains under the conveyor boom, 
 continuing to receive coal until it is fully loaded. When 
 loaded, the shuttle car leaves the face area to empty its 
 load, and is replaced by a second and third empty shuttle 
 car, which is positioned under the conveyor boom to re- 
 ceive more coal. When the cutting process is complete, 
 i.e., when the face has been advanced to the point where 
 the operator is under the last row of roof bolts, the con- 
 tinuous mining machine and the shuttle car leaves the face 
 area to allow the roof bolting machine access to the newly 
 cut area. The roof bolter is positioned under the unbolted 
 roof, driUing holes and inserting bolts at the required 
 spacing. 
 
 The job of locail navigation is the guidance of mobile 
 mining equipment in the local (immediate) area, excluding 
 the face area, i.e., guidance around corners, through en- 
 tries, and around obstacles. For example, a continuous 
 mining machine is constantly maneuvering among face 
 areas and is guided through various connecting passages. 
 A loaded shuttle car is guided through each entry as it 
 travels a predetermined path from the face to the dump 
 point. 
 
 The job of global navigation is the navigation of mobile 
 mining equipment between various points in the mine, i.e., 
 path planning and following from one work area to 
 another. An operator of a mine haulage vehicle, e.g., must 
 plan and follow a path from the material loading point to 
 various work areas currently requiring the materials. 
 
 NAVIGATIONAL TASKS OF CONTINUOUS 
 MINING MACHINE 
 
 The focus of the Bureau's research into autonomous 
 navigation in mines is the navigational tasks of the contin- 
 uous mining machine. An operator of a continuous miner 
 performs all three different types of navigational tasks in 
 mines; face, local, and global. Face navigation, although 
 specific to the equipment and the task being performed, 
 requires the most accuracy and reliabihty for a continuous 
 mining machine. Local and global navigational tasks are 
 similarly performed by operators of all mobile mining 
 equipment. Therefore, any navigational techniques proven 
 to work for the continuous mining machine can be applied 
 to other mobile mining machinery. 
 
 Face Navigation of Continuous Mining Machine 
 
 An operator's objective during face navigation is to con- 
 trol the position and heading of the continuous miner dur- 
 ing the cutting process in order that coal is removed 
 according to the mine plan. Although many different coal 
 removal processes exist, i.e., full-face mining, longwall 
 mining, shortwall mining, etc., this research will highlight 
 the guidance for a two-pass continuous mining machine 
 performing straight-cuts and crosscuts in room-and-pillar 
 mining. Room-and-pillar mining is a widely used type of 
 mining. A typical section is shown in figure 1. Longwall 
 mining has become more prevalent in recent years; how- 
 ever, room-and-pillar mining is still used for development 
 of both longwall and shortwall sections. Additionally, the 
 
 4 
 
Figure 1.— Typical room-and-pillar mining section. 
 
 navigation of a two-pass machine, as opposed to a full-face 
 miner, longwcdl machine, etc., includes much more maneu- 
 vering and guidance. Therefore, many of the navigational 
 issues will be easily applicable to other types of mining 
 processes. 
 
 The two most significant issues of face navigation are 
 the alignment of the continuous mining machine at the 
 face, and the guidance of the machine during the cutting 
 process. The following discussion of the methods used by 
 continuous miner operators to perform these tasks indi- 
 cates the navigationail information and guidance techniques 
 required at the face. 
 
 Alignment at Face 
 
 The first task of a continuous miner operator is plac- 
 ing the mining machine in the correct position and heading 
 at the face to begin the cutting process. The operator 
 
 requires an indication of the position, heading, and dimen- 
 sions of the desired cut. The position and heading of the 
 cut is precisely mapped on a mine plan. Surveying marks, 
 a laser spot, or a vertical laser guideline are located in the 
 mine according to the plan, indicating the centerline of the 
 entry. The continuous miner operator observes the loca- 
 tion of the centerline indication to judge the correct posi- 
 tion and heading for the machine at the face. 
 
 The operator trams the machine up to the face while 
 avoiding collisions with the ribs, roof, or obstacles. The 
 operator is aware of the position of all exterior points of 
 the machine relative to any of these objects. The brattice 
 cloth is the assumed method of ventilation control, and is 
 also an obstacle at the face which must be avoided by the 
 continuous miner. Therefore, the operator aligns the con- 
 tinuous miner at a slight angle relative to the heading of 
 the desired cut. 
 
Guidance During Cutting Sequence 
 
 During the execution of the cutting process, the 
 operator continues to observe the location of the surveying 
 marks, laser guideline, and/or the orientation of the ribs, 
 and judges the position of the machine relative to the 
 intended centerline of the cut. However, these actions are 
 not a precise method of guiding the continuous mining 
 machine, hence the accuracy of the cut is dependent on the 
 operator's skill and experience. 
 
 The maneuvering sequences for a straight cut and a 
 crosscut are shown in figures 2 and 3. The continuous 
 miner operator makes each cut to the appropriate height, 
 width, and depth, with the centerline of the cut being as 
 close to the centerline indication as possible. In practice, 
 the dimensions of the desired cut are determined from the 
 characteristics of the surrounding strata. However, for this 
 research, a cut is assumed to be a 20-ft wide by 20-ft deep 
 advancement of the face while remaining within the coal 
 seam. 
 
 The ability of the continuous mining machine to follow 
 the vertical undulations of the coal seam is not within the 
 scope of this project. Another current research effort at 
 the Bureau is investigating coal interface detecting sensors 
 for vertical guidance of coal extraction machines. The 
 horizontal alignment and guidance of the continuous min- 
 ing machine are the only tasks of face navigation presently 
 within the scope of the Bureau's autonomous navigation 
 research program. 
 
 Local Navigation of Continuous Mining Machine 
 
 The navigational tasks of a continuous mining machine 
 in the local area consists of the guidance of the continuous 
 miner through entries and around corners, while avoiding 
 collisions with ribs, workers, equipment, or other obstacles. 
 The following discussion of the methods used by continu- 
 ous miner operators to perform these tasks indicates the 
 navigational information and guidance techniques required 
 for autonomous navigation in the local area of a mine. 
 
 Tramming Through Entry 
 
 Before tramming through an entry, an operator ob- 
 serves the area in the direction of travel to find a travers- 
 able path. If a path is located, and it is certain that the 
 inclination is not too great and the terrain not too rough, 
 the operator guides the continuous miner along the path. 
 
 While traveling along the path, the operator is con- 
 stantly aware of the position of all exterior points of the 
 machine relative to the roof, ribs, and obstacles. The 
 operator also watches in the direction of travel of the 
 mining machine. People, supplies, trailing cables, equip- 
 ment, or other mobile vehicles are common obstacles in a 
 mine which must be avoided. Even small obstacles, such 
 as small rocks, can hinder the navigation or even damage 
 the large mining machine if encountered. To lessen navi- 
 gational difficulties caused by small rocks, the operator 
 
 may lower the gathering head, and clean the path while 
 tramming. 
 
 Turning Corner 
 
 Corner navigation in mines is a tricky maneuver owing 
 to the limited mobility of the large continuous miner in 
 narrow entries. The position of the corner is observed, 
 and the operator elects a proper point at which to initiate 
 the turn. The operator performs a series of forward and 
 backward maneuvers to execute the turn in the limited 
 space. While maneuvering, the operator judges the track 
 slippage to faciUtate control of the turn and turning radius 
 of the machine. 
 
 Limited visibility is another difficulty in corner navi- 
 gation. An operator makes certain that people or other 
 mobile vehicle operators are aware of the machine, and 
 that the operator is aware of their presence, to avoid in- 
 juries or collisions. 
 
 Global Navigation of Continuous Mining Machine 
 
 The navigation of the continuous mining machine in the 
 global area consists of planning and following a path 
 among work areas in the mine. This navigation requires 
 the abiHty, given a map of the mine, to generate a path to 
 a predetermined destination, follow that path until the des- 
 tination is reached, and update the map of the mine when 
 necessary. The following discussion of the methods used 
 by continuous miner operators to perform these tasks indi- 
 cates the navigational information and guidance techniques 
 necessary for autonomous navigation in the global area of 
 a mine. 
 
 Generate Path 
 
 Given a destination, a continuous miner operator gener- 
 ates a path. The operator also determines an alternate 
 path when an entry through which the operator intended 
 to travel is not traversable or blocked. 
 
 To generate paths, the operator has a knowledge of the 
 common paths of travel in the particular mine. In addi- 
 tion, the operator may know what areas are dangerous, 
 and entries that have a difficult terrain for travel. This 
 knowledge could be provided to the operator in the form 
 of visual markers placed for guidance, a memory of past 
 experience of the paths, or a map of the mine. 
 
 Follow Path 
 
 The operator travels along the route, one entry at a 
 time. The operator navigates through entries until where 
 to turn is recognizable. If the operator is unfamiliar v^ith 
 the route, the operator may keep track of the position 
 relative to the map by counting crosscuts, observing mark- 
 ers on the map or in the entry, or by keeping track of dis- 
 tance traveled and turns made. 
 
I'iim fr^ii lin iiiiiiiaiiiittiiirii liiiiriffiiiii iti 
 
 :: 
 
 t 
 
 Figure 2.— Maneuvering sequence for straight-cut in room- 
 and-pillar mining with a brattice cloth in the face area. 
 
 Figure 3.— IVIaneuvering sequence for crosscut in room-and- 
 pillar mining with a brattice cloth in the face area. 
 
Update Map 
 
 The operator records any permanently blocked passage- 
 ways on the map of the mine (or by memory), so they will 
 be avoided in futm-e travel. New entries that are cut by 
 the continuous miner during face navigation are likewise 
 recorded, to be used as new paths of global travel, or 
 indicators of globsJ position. 
 
 PREVIOUS RESEARCH AND ATTEMPTS IN REMOTE 
 
 AND AUTONOMOUS NAVIGATION OF 
 
 CONTINUOUS MINING MACHINES 
 
 Previous research into remote and automatic guidance 
 of continuous mining machines provides valuable infor- 
 mation to the current effort in autonomous navigation in 
 mines. The following summary of research attempts is not 
 complete, but gives some of the previous research that 
 significantly affected the direction of the current autono- 
 mous navigation project. 
 
 Laser Alignment Sensor for Continuous 
 Mining Machines 
 
 An entry alignment sensor to provide lateral guidcmce 
 for a continuous mining machine was developed and tested 
 in an operating mine by Bendix under a Bureau contract.^ 
 The purpose of the sensor is to provide a guideline for 
 entry heading during remote or automatic control of the 
 cutting process. It consists of a scanning laser source to 
 project a vertical guideline, and a sensor unit containing an 
 array of photodetectors to analyze the heading and lateral 
 displacement of the received beam. 
 
 The laser source, located in the entry, projects a beam 
 parallel to the centerline of the desired cut. The sensor 
 imit, located on the continuous mining machine, contains 
 seven fresnel lenses that focus the laser guideline onto 
 split photodetectors. The signals from the detectors are 
 balanced when the machine is correctly positioned parallel 
 to the laser guideline. An unbalanced signal is propor- 
 tional to the heading displacement of the machine. Since 
 an array of these split detectors are employed, the latercd 
 displacement relative to the centerline is also determined. 
 This information of heading and lateral displacement is 
 then indicated on a display outby the machine, and used in 
 remote guidance of the continuous mining machine. 
 
 The employment of the laser source and sensor unit 
 system proved to be a moderately successful means of 
 remotely guiding a continuous mining machine in the 
 mining environment. Some difficulties would have to be 
 overcome if the system were employed for autonomous 
 guidance of a continuous miner. 
 
 Bowser, M. L. Laser Alinement Sensor for Continuous-Mining 
 Machines. BuMines RI 8134, 1976, 10 pp. 
 
 The major difficulty occurred when the beam was 
 broken by obstacles. If the becun was not received by the 
 sensor unit, the most recent reading was assumed until the 
 beam was received again. Obviously, the reliability of the 
 system was reduced, and if the beam was interrupted for 
 a long period of time, the operation of the machine had to 
 be suspended until the beam was received. 
 
 Shuttle cars were the most frequent cause of beam in- 
 terference. At other times, the line curtain interfered with 
 the beam, and had to be moved or opened. And finally, 
 undulations in the coal secun over a great distance were a 
 cause of the loss of the line-of-sight. 
 
 In summary, the Iciser alignment sensor system could 
 become a successful means of autonomous guidance, if the 
 laser were mounted in a position where the line-of-sight 
 could be maintained. Also, the ability of the laser to ad- 
 vance autonomously in the entry as the continuous mining 
 machine advances would eliminate the problem of undula- 
 tions in the seam causing a loss of the line-of-sight. 
 
 Assessment of Optic Guidance 
 and Supervisory Methods 
 
 An evaluation of optical guidance systems for auto- 
 mated face navigation was prepared by MBAssociates un- 
 der Bureau contract J0155084, Assessment of Optic Guid- 
 ance, Control, and Supervisory Methods. The evaluation 
 includes information on the employment of optical systems 
 for the lateral alignment of a continuous miner. 
 
 MBAssociates determined the characteristics of lateral 
 alignment of a continuous mining machine, which affect 
 the performance of optical guidance systems. They are 
 listed as follows: 
 
 1. Any optical path along the entry outby the continu- 
 ous miner will be occluded from time to time. 
 
 2. The percentage of occluded time increases as the 
 seam height decreases. 
 
 3. Face area illumination standards will strongly influ- 
 ence the ambient light level. As a note, these standards 
 would not be required in a fully automated face area. 
 
 4. The continuous miner can be expected to make 
 gross movements vertically and horizontally (several inches 
 to a foot) while cutting. 
 
 5. Any equipment mounted on top of the continuous 
 miner is vulnerable to damage and contamination from 
 roof falls, contact with the roof, and coal kicked back from 
 the face. 
 
 6. The continuous miner follows the coal seam, there- 
 fore the entry will rise and fall along its length (line-of- 
 sight is not reliable). 
 
 Two basic lateral alignment systems using a vertical 
 laser guidelines were investigated by MBAssociates. The 
 first system is similar to the laser alignment sensor 
 
 1 
 
developed by Bendix. The second system uses the same 
 concept, but exchanges the position of the laser source and 
 sensing unit. 
 
 The first system, with the laser source mounted in the 
 entry, was shown to have three disadvantages. First, a 
 rehable system would require a large sensing imit. Other- 
 wise, the laser guideline would be lost during the move- 
 ments of the continuous mining machine during the cutting 
 sequence, and the machine would shut down. Secondly, 
 two laser guidelines would be required since two lifts are 
 taken for each straight-cut. And, finally, the sepcu^ation of 
 the sensing unit from the continuous miner would neces- 
 sitate a data link to the continuous mining machine where 
 none currently existed. 
 
 The second system had the laser source mounted on the 
 machine, and the sensing unit in the entry. The vertical 
 laser guideline would be swept horizontally across the 
 entry, analogous to a rastor scan, and the system would 
 require only one laser guideline and one sensor. However, 
 MBAssociates found disadvantages with the system in the 
 extra complications with the horizontal sccuining hardware, 
 and the separation of the sensing unit from the continuous 
 miner. 
 
 Another possible optical guidzmce system, investigated 
 by MBAssociates, was a low-level pattern recognition sys- 
 tem. Three narrow-band hght sources would be mounted 
 on the continuous miner. A television camera would be 
 positioned in the entry with its optical axis parallel to the 
 centerUne of the desired cut. The position of the lights in 
 the image would determine the lateral position and head- 
 ing of the continuous miner. 
 
 MBAssociates found two disadvantages with this system. 
 The first disadvantage is the number of lights required for 
 position and heading guidance. Three lights would be 
 required to indicate the lateral displacement and heading 
 of the continuous miner. The more Ughts used, the more 
 chance that one will be blocked by an object. And again, 
 a separation exists between the sensing unit and the con- 
 tinuous miner. 
 
 For all the optical systems considered, the major diffi- 
 culty was maintaining the line-of-sight between the sensor 
 and source. Shuttle cars proved to be the major obstruc- 
 tion. The conclusion made by MBAssociates is that optical 
 guidance system, such as those described, may not be sat- 
 isfactory, especially with shuttle car haulage in the face 
 area. 
 
 Preliminary Design of Automated 
 
 Continuous Mining Machine 
 
 With Remote Override 
 
 Ingersoll-Rand Resec^ch under Bureau contract 
 H0242051, Preliminary Design of an Automated Continu- 
 ous Mining Machine With Remote Override, contains the 
 preliminary specifications for three automated continuous 
 mining machines. Techniques are proposed for determin- 
 ing heading displacement given an initial reference 
 heading. 
 
 The use of a gyroscope was suggested to provide 
 heading displacement for the continuous miner. The de- 
 sired heading of a cut would be initially given, and any 
 small changes in heading could then be detected by the 
 gyroscope. 
 
 A gyroscope has an inherent drift in heading over time. 
 Therefore, it requires realignment to some reference at 
 regular intervals to maintiiin accurate readings. Ingersoll- 
 Rand proposed that the gyroscope be realigned to true 
 north at regular intervals by stopping the machine and per- 
 forming a gyrocompassing procedure. They determined 
 that the 10-min gyrocompassing procedure would be re- 
 peated every 4 to 8 h. Since the specification of drift on 
 the gyroscope to be used was 0.1°/h the drift error was 
 tolerable. 
 
 In today's market, however, a gyroscope with such a low 
 drift costs $200,000. Reasonably priced gyroscopes, such 
 as those used on miUtary vehicles today, have a much 
 higher drift and would require the gyrocompassing proce- 
 dure much too often. A gyroscope only provides a reading 
 of change in heading, and provides no data on translational 
 movements. It is not economically feasible to employ such 
 a costly device when additional sensory systems are neces- 
 sary to provide full guidance information. 
 
 Ingersoll-Rand suggested the use of a magnetic flux- 
 gate sensor to provide relative heading information. This 
 electronic sensor directly measures the Earth's magnetic 
 field to determine heading. However, owing to the mag- 
 netic interference present on the continuous miner, the 
 sensor's reading of absolute heading relative to true north 
 is inaccurate, and cannot be used as a constant reference 
 for entry heading. Nonetheless, Ingersoll-Rand suggested 
 that a magnetic flux-gate sensor could possibly be used in 
 a manner similar to the gyroscope and would provide reh- 
 able measurements of small changes in heading, if the 
 continuous miner were initially aligned to some known 
 heading. 
 
 CONCLUSIONS OF RESEARCH 
 
 The previous discussions of the tasks of autonomous 
 navigation in mines, and the research into sensors and 
 navigational techniques (see appendix) leads to the follow- 
 ing conclusions for the tasks of face, local, and global navi- 
 gation of a continuous mining machine. 
 
 Face Navigation 
 
 Face navigation of a continuous mining machine con- 
 sists of the task of controlling the position and heading of 
 the continuous miner during the cutting process so that 
 coal is removed according to the mine plan. This task re- 
 quires that the continuous miner be provided with a refer- 
 ence indicating the position and heading of the desired cut, 
 along with precise guidance abilities. 
 
Reference for Desired Cut 
 
 Guidance During Cutting Sequence 
 
 Surveying marks, laser guidelines, or laser spots, the 
 references currently used by continuous miner operators, 
 cannot be relied upon as a continual indication of entry 
 centerline unless they can advance and remain in close 
 proximity to the working face and the continuous miner. 
 In that case, a surveyor is required to place marks or 
 advance the laser further into the entry as the machine 
 advances. The surveyor's presence at the face would be 
 frequently required owing to undulations in the seam, and 
 the limitations in range of the guideline or the sensor to 
 access the position of the surveyor's mark. This solution 
 defeats the goal of removing workers from the working 
 face. 
 
 Compasses cannot be used as a constant reference for 
 entry heading, owing to the magnetic interference caused 
 by the large iron continuous mining machine. 
 
 Laser-based guidance systems show promise to provide 
 precise, reliable guidance for continuous mining machines 
 when the Hne-of-sight is maintained. The major short- 
 coming in previous laser guidance systems was the position 
 of the laser or sensor in the entry. It was located in the 
 center of the entry outby the continuous mining machine, 
 causing a loss of line-of-sight by the shuttle car. If the 
 laser or sensor was positioned on the sides or top of the 
 entry and in close proximity to the continuous mining 
 machine, this problem could most likely be eliminated. 
 
 The sensory system employed to view the reference and 
 determine the position of the continuous miner relative to 
 the desired cut, should not fail if one sensor is not func- 
 tioning properly. For example, the laser alignment sensor 
 developed by Bendix used a single user guideline to indi- 
 cate entry position and heading. The system is totally 
 dependent on a single device. If the line-of-sight was not 
 maintained, the operation of the machine would have to be 
 suspended. To avoid this situation, relevant data should 
 be gathered by multiple sensors and, if any sensors fail, 
 other sensors will continue to provide adequate guidance 
 information. 
 
 In conclusion, the reference of the desired cut could be 
 obtained by employing a multiple laser system mounted 
 on a structure located near the face, and a sensor unit on 
 the machine. The most reliable guidance would be ob- 
 tained if the lasers were located on the sides or top of the 
 entry, to best maintain the line-of-sight. The structure 
 could initially be positioned at the face indicating the posi- 
 tion and heading of the desired cut, providing the machine 
 with the feedback information necessary for guidance at 
 the face. When the machine completed a cutting cycle, it 
 would stop, and the structure, if provided with some meth- 
 od of self-locomotion, could advance in the entry using the 
 stationary machine as a reference. 
 
 Adequate guidance while advancing the face can be ob- 
 tained by viewing the reference for the desired cut, and 
 determining the relative position of the continuous miner 
 with respect to that reference. 
 
 In addition, the continuous miner must be able to navi- 
 gate for a short period of time when it may not see the 
 reference. Therefore, other sensory systems which do not 
 require sight of the reference will be used to provide infor- 
 mation when the referencing sensors fail or cue in error. 
 
 Inertial reference sensors on the continuous miner 
 could be used to maintain an accurate short-term knowl- 
 edge of position and heading from a known initial location, 
 thereby requiring the reference for the desired cut at reg- 
 ular intervals rather than continually. At times when re- 
 dundant, valid data are obtained from these multiple sen- 
 sory systems, they will be combined to provide a greater 
 and more reliable awareness of the position and heading 
 of the continuous miner. 
 
 Local Navigation 
 
 Local navigation of a continuous mining machine con- 
 sists of guidance and collision avoidance through entries 
 and around corners. Adequate guidance and collision 
 avoidance can be obtained by viewing the area in the 
 direction of travel to find a clear path, and following the 
 path while recognizing the presence and position of objects 
 near the extremities of the machine to avoid collisions. 
 The sensors and navigational techniques employed by vari- 
 ous autonomous and remote vehicles in other local envi- 
 ronments have been reviewed (see appendix), and many 
 can be applied to the local navigation in the mine. 
 
 Find Patli 
 
 The machine must be able to see in the direction of 
 travel to determine a path. Scanning range sensors, or 
 sensors with a wide field-of-view are necessary to detest 
 objects in the continuous miner's path. The range of the 
 sensors must allow enough time for the continuous miner 
 to take action to avoid an object in its path. 
 
 The mounting of forward looking sensors on the contin- 
 uous miner will be more difficult than on other mobile 
 vehicles. A raised cutting head obscures frontal vision. 
 Also, any sensors mounted toward the front of the contin- 
 uous miner will experience maximum vibration and dust 
 obscuration during the cutting cycle. Fragile imaging or 
 scanning devices are not recommended since they will not 
 survive this harsh treatment. 
 
Follow Path 
 
 SUMMARY 
 
 Ranging sensors, with their data processed with 
 environmental modeUng techniques, provide a useful 
 physical model of the immediate environment for local 
 path following. To be specific, the position of ribs, 
 crosscuts, or obstacles can be monitored with continually 
 updated data from ranging sensors. The continuous miner 
 is then provided with the information necessary for 
 obstacle avoidance, wall following and guidance when 
 traveling through entries and around corners in a mine. 
 
 Reliable short-range collision avoidance of the large 
 continuous mining machine in the close quarters of a mine 
 is critical. As stated in the appendix, a reliable time-of- 
 flight sensor with a wide field-of-view is the best ranging 
 sensor for collision avoidance. Ultrasonic ranging sensors, 
 commonly employed for obstacle avoidance on other 
 autonomous vehicles, have a wide field-of-view, and show 
 promise for reliability in the mine environment.^ 
 
 Global Navigation 
 
 Global navigation is the navigation among various 
 points in the mine. The continuous miner requires a pre- 
 programmed globed mine map along with the ability to 
 generate and follow a path on the mine map. The map 
 includes common travel routes, markers that may indicate 
 global position, ventilation that may inhibit travel, or ter- 
 rciin that cemnot be traversed. 
 
 Generate Path 
 
 The machine must be able to process the mine map and 
 determine a path. Initially, the mining machine will be 
 given its ciu-rent position relative to the mine map, and a 
 final destination position. A few simple heuristics will 
 provide the global path planning necessary in a mine (see 
 appendix). 
 
 Follow Path 
 
 The continuous miner must be able to follow its chosen 
 path. The current technology of inertial reference sensors 
 does not provide long-term global positioning information. 
 Markers could be made available at regular intervals; how- 
 ever, these markers may get torn down by machinery or 
 obscured by dust. The most promising method of global 
 path following in a mine is simply to break the path down 
 into a series of straight line segments, and count the cross- 
 cuts passed and turns made to keep track of position rela- 
 tive to the map. This information can be obtained from 
 the environmental model developed for local navigation. 
 
 ^Bauzil, G., M. Droit, and P. Ribes. A Navigation Sub-System Using 
 Ultrasonic Sensors for the Mobile Robot Hilare. Paper in Proceedings 
 of First International Conference on Robot Vision and Sensory Con- 
 trols, Stratford-upon-Avon, UK, Apr. 1-3, 1981, IFS, pp. 47-58. 
 
 Oppenheim, I., and W. Whittaker. Demonstration of Robotic Map- 
 ping of Mine Spaces. Eng. Constr. Rob. Lab., Carnegie-Mellon Univ., 
 Pittsburgh, PA, Apr. 1985, 58 pp. 
 
 The Bureau is proposing to develop a navigational guid- 
 ance system to provide a continuous mining machine with 
 the ability to navigate autonomously in a mine. Three 
 types of navigation were identified: face, local, and global. 
 
 Autonomous face navigation requires that the contin- 
 uous miner be provided with an indication of the position 
 and heading of the desired cut, and the abUity to monitor 
 its movements at the face. A multiple laser-based guid- 
 ance system mounted on a mobile reference structure is 
 suggested. The system would provide the machine with 
 the feedback information necessary to maneuver through 
 the cutting sequences.. After a cutting sequence is com- 
 plete, the structure could advance, using the stationary 
 machine as a reference and maintaining an accurate indi- 
 cation of entry position and heading. Inertial reference 
 sensors could also be incorporated in this system to pro- 
 vide short-term knowledge of the machine's position when 
 the laser system may fail. 
 
 Autonomous local navigation consists of the tasks of 
 guidance and collision avoidance through entries and 
 around corners. The machine must be provided with an 
 ability to see in the direction of travel, and a knowledge of 
 the position of the ribs and any obstacles. Ranging sen- 
 sors, with their data processed with environmental model- 
 ing techniques, provide a useful physical model of the 
 immediate environment for local path following. Ultra- 
 sonic ranging sensors, with their wide field-of-view, are 
 suggested to obtain the data. 
 
 Autonomous global navigation consists of the task of 
 planning and following paths from one area in a mine to 
 another. The machine must be provided with a prepro- 
 grammed map of the mine, along with the ability to gen- 
 erate and follow the planned path. The most promising 
 method of global path following in a mine is simply to 
 break the path down into a series of straight-line segments, 
 and to count the crosscuts passed and turns made to keep 
 track of position relative to the map. 
 
 Of the three types of navigation in mines identified, the 
 navigation of a continuous mining machine at the face is 
 one of the more dangerous jobs in the mine. The high 
 levels of noise and respirable dust, common at a working 
 face, pose a personal health risk. Also, during the cutting 
 process, the continuous miner operator is in close proxi- 
 mity to unsupported roof. In the close confinement of the 
 cab, the operator may lean out from the protective cover 
 to gain a better view of the area. The operator is then ex- 
 posed to the potential injury from falling rock. Also, the 
 operator is endangered by possible explosions due to the 
 higher concentrations of methane and dust, common at a 
 working face. Consequently, many hazardous situations 
 could be eliminated by automating the tasks performed by 
 the continuous miner operator in the face area. Although 
 automating these tasks alone will not eliminate the neces- 
 sity for workers in mines, it will permit the immediate 
 relocation of continuous miner operators from a highly 
 dangerous environment to less hazardous areas, thus in- 
 creasing the overall safety of the mine worker. 
 
10 
 
 SENSORY SYSTEM, DATA MANIPULATION, AND CONTROL PROGRAMMING 
 
 INTRODUCTION 
 
 A decision to concentrate initial attempts on the tasks 
 of autonomous face navigation has been made. This sec- 
 tion outlines a proposed solution for autonomous face 
 navigation of a continuous mining machine including the 
 employment of a mobile roof support as a reference for 
 guidance in the face area, a sensory system to provide 
 navigation information, and algorithms for data manipu- 
 lation and navigational control. 
 
 AUTONOMOUS FACE NAVIGATION OF 
 CONTINUOUS MINING MACHINE 
 
 positions and headings necessary to execute a straight-cut 
 and a crosscut in room-and-pillar type mining. The cutting 
 sequences are as shown in figures 2 and 3. These cuts 
 must also be executed according to the mine plan, to main- 
 tain a safe amd productive mine. This task requires that 
 the continuous miner be provided with a physical reference 
 in the face area indicating the position and heading of the 
 desired cut, and a navigational guidamce system consisting 
 of a method to access the reference, to obtain a knowledge 
 of its physical surroundings, and to control and monitor its 
 maneuvers in the face area. 
 
 Reference for Desired Cut 
 
 The current goal of the project is to develop a naviga- 
 tional guidance system to provide a continuous mining 
 machine the ability to navigate autonomously through the 
 
 The proposed solution is to employ a mobile control 
 structure (MCS) as a reference for guidance in the face 
 The MCS is currently under development at the 
 
 area. 
 
 6 
 
 Figure 4.— Mobile roof support (MRS). 
 
11 
 
 Bureau. It will provide a control center for all face oper- 
 ations in a mine, and will conceivably serve many other 
 functions, such as bolting the roof, providing a source of 
 ventilation, performing methane checks, etc. However, 
 when precisely positioned in the face area indicating both 
 the position and heading of the desired cut, the MCS could 
 provide a fixed, stable reference to guide all face navi- 
 gation. The physical design of the MCS will be patterned 
 after a mobile roof support (MRS), a commercially avail- 
 able device designed to provide a temporary means of 
 additional or alternative roof support (fig. 4). It consists 
 of laterally spaced bars on four hydrauUcally powered 
 wheels and four vertical hydraulic rams. The MRS is 
 guided into position and the bars are forced against the 
 roof by the rams. 
 
 machine would refer to the position of the MCS as a refer- 
 ence for alignment (fig. 5). During the cutting process, the 
 continuous miner would continue to update its position 
 and heading relative to the reference. The continuous 
 miner would always be in close proximity to the MCS and 
 could easily and frequently update its position and heading 
 during both a straight-cut and crosscut sequence. 
 
 After a full straight-cut is completed, the MCS would 
 automatically advance 20 ft in the entry. It would use the 
 position of the now stationary continuous mining machine 
 as a reference, thereby maintaining a constant heading to 
 guide the next cut. Also, it would have the abiUty to pivot 
 90° and provide the reference for a crosscut. 
 
 Benefits of Guidance From MCS 
 
 Description of Guidance from MCS 
 
 Initially, the MCS would be guided into place by a 
 human using precise surveying equipment. As it is maneu- 
 vered underneath the MCS, the continuous mining 
 
 The MCS would provide a more protective environment 
 for mounting of delicate sensors and equipment than the 
 continuous miner. It would experience much less move- 
 ment and vibration. The MCS and continuous mining ma- 
 chine would function together via a communication link. 
 
 Figure 5.— Continuous miner advancing the face under IVICS. 
 
12 
 
 and gather and share environmental information. There- 
 fore, the expensive and more fragile sensory systems (e.g., 
 optical scanning devices) would be placed on the MCS, 
 and the cheap passive targets and more rugged sensors 
 would be placed on the continuous mining machine. 
 
 The use of the MCS could eUminate more human inter- 
 action in the face area. The MCS has the ability to func- 
 tion not only as a reference for the continuous miner, but 
 likewise as a reference for other mobile mining equipment, 
 such as haulage vehicles and roof-bolting machines. At 
 this time, their guidance is not the subject of research; 
 however, if the MCS guidcuice techniques are successful 
 for the continuous mining machine, then they could be 
 easily applied to these other machines. Also, the MCS 
 could provide a method of automatic ventilation, as re- 
 quired by law, possibly by supporting and advancing a line 
 curtain or by carrying an auxiliary fan or tubing. The need 
 for human workers to be present at the working face 
 would then be considerably reduced or eliminated. 
 
 The use of the MCS could speed up the advancement 
 of the face. It provides a means of roof support for an 
 area past the last row of roof-bolts at the face. The 
 continuous miner would then be safely able to advance the 
 face further, thereby eliminating the number of place 
 changes necessary to allow roof-bolting equipment access 
 to the unsupported roof. 
 
 Disadvantages of MCS Reference System 
 
 The disadvantage in employing a MCS as a reference is 
 that it is an additional mining vehicle to automate. The 
 MCS would require its own autonomous positioning and 
 guidance system, including sensors, processing capabilities, 
 and motion control system. Obviously, the cost-benefit 
 ratio is high when considering it only as a reference for the 
 guidance of the continuous mining machine. However, in 
 concept, the MCS would eventually be employed as the 
 central guidance system for all vehicles in the face area. 
 And, therefore, if one considers the resulting increase in 
 production and decrease in labor force, the cost-benefit 
 ratio would decrease. 
 
 Navigational Guidance System 
 
 Figure 6 is the proposed hardware-software systems and 
 information transfer for a navigational guidance system 
 providing autonomous face navigation for a continuous 
 mining machine. 
 
 Sensory Solution 
 
 The previous section of this report indicated specific 
 sensors, which show promise for the navigation of remote- 
 controlled and autonomous mining vehicles. Laser-guided 
 mining vehicles proved successful when the line-of-sight is 
 
 maintained."* Also, another previous research attempt into 
 the automation of continuous mining machines concluded 
 that gyroscopes providing an inertial reference, with a 
 periodic method to reference a known heading, could ef- 
 fectively be used as a guidance sensor. Ultrasonic ranging 
 systems have been demonstrated to be reUable for naviga- 
 tion in mines,^ and for coUision avoidzmce on many mobile 
 robots.* Also, the report indicated thai no one sensor 
 should be reUed upon for guidamce information, but many 
 sensors should be employed. Multiple sensors, with their 
 data fused together, provide a more accurate and rehable 
 aw£U"eness of the environment. 
 
 As a result of the previous research, the following sen- 
 sors have been selected. A gyroscope and a laser system 
 has been chosen to be incorporated together for the pri- 
 mary guidance information. A laser system mounted on 
 the MCS will provide a direct reading of heading and posi- 
 tion relative to the desired cut. A gyroscope mounted on 
 the continuous miner will provide an inertial reference for 
 heading, from the last heading determined from the laser 
 system. An ultrasonic ranging system will be employed to 
 obtain a knowledge of the physical surroundings. In addi- 
 tion, a flux-gate heading sensor will be tested to determine 
 its feasibility as a replacement or aid to the gyroscope. 
 And, finally, inclinometers wUl be employed to provide a 
 direct measurement of the pitch and roll of the continuous 
 mining machine. 
 
 The following is a description and implementation of 
 the sensors that will be used to provide data for 
 autonomous control of face navigation of a continuous 
 mining machine. 
 
 Laser-based Scanning System 
 
 Lasernet,^ a laser-based optical scanning system devel- 
 oped by Namco Controls, Mentor, OH, will be employed 
 to access the reference and determine the position and 
 heading of the machine relative to the desired cut. This 
 relevant information will be gathered by multiple laser 
 units, and if any units fail, others will continue to provide 
 adequate guidance information. 
 
 The Lasernet system contains a rotating mirror that 
 sweeps a horizontal beam of laser light across a 90° field- 
 of-view at a constant angular velocity. When the beam 
 encounters a special 12-in tall cylindrical retroreflective 
 target, the return beam is detected by a photodetector. 
 This system can analyze information about the return 
 beam and report the angle to the target. It is capable of 
 determining the angular location of up to eight retroreflec- 
 tive targets located within the planar and angular range of 
 the scan. 
 
 Work cited in footnote 2. 
 ^Second work cited in footnote 3. 
 *First work cited in footnote 3. 
 
 ^Reference to specific products does not imply endorsement by the 
 Bureau of Mines. 
 
13 
 
 Sensor data 
 
 (Engr. units) 
 
 Monitoring 
 systems 
 
 Raw data 
 
 from sensors 
 
 Integration 
 system 
 
 Local 
 
 model 
 
 Navigational 
 
 task 
 
 scheduler 
 
 Current Goal Local 
 state, state, model 
 
 Machine response (Ax,Ay,A^) 
 
 Laser 
 
 scanning 
 
 system 
 
 Gyroscope 
 
 compass, 
 
 inclinometers 
 
 Ultrasonic 
 ranging 
 system 
 
 Action controls 
 
 to actuators 
 
 Figure 6.— Block diagram of the hardware-software systems and information transfer for proposed navigational guidance system. 
 
 Mtiltiple laser-scanning devices will be mounted on the 
 MCS, and multiple targets wiU be mounted at a known 
 separation on the continuous mining machine. A data link 
 between the continuous miner and the MCS will provide 
 the continuous miner with the feedback information neces- 
 sary to determine its position and heading relative to the 
 desired cut for aUgmnent and guidance at the face. 
 
 Determining Position and Heading Using 
 l-aser-Based Scanning System 
 
 The position and heading of the continuous miner can 
 be determined by two methods. 
 
 Method 1: Given the angular position of two targets (A 
 and B) of known position on the continuous miner relative 
 to two laser-scanning devices (LI, L2) (fig. 7). 
 
 The position of each target relative to LI (the desig- 
 nated origin) can be determined using trigonometry. As 
 shown in figure 8 for both targets A and B, the separation 
 between the laser-scanning devices represents the length of 
 the baseline, and each target position represents the vertex 
 of a tricmgle. 
 
 Since the separation between laser devices (dl2, dl2) 
 is knovm, and the angles to the targets (6^^, 9^ and ^b„ 
 9^2) are given, the length of the side (dAl, dBl) can be 
 determined with triangulation. The position of each target 
 relative to LI can then be determined by projecting the 
 sides of the triangles onto the x and y dods. And, finally, 
 the heading (h) of the targets relative to the centerhne of 
 the cut (y axis) can be calculated from the two positions, 
 
 h - tan" 
 
 ^B 
 
 Ya - Yb 
 
 Method 2: Given the angular position of three targets 
 (D, E, and F) relative to one laser-scanning device (L3) 
 
 (fig- 9). 
 
 The position of the targets relative to L3 (the desig- 
 nated origin) can be determined using trigonometry. As 
 shown in figure 10, three triangles are formed by repre- 
 senting the position of L3 as a common vertex, and the 
 various target separations as the baselines of the triangles. 
 
14 
 
 A(Xa,Ya) 
 
 ■>X 
 
 I B(Xg,YQ) 
 
 Figure 7.— Method 1 for determining position and heading of 
 continuous miner relative to MCS. Angular position of two tar- 
 gets (A,B) on the continuous miner relative to two laser scanning 
 devices (L1,L2) is known. 
 
 Fiqure 8.— Mathematcal representation of method 1 for posi- 
 tion determination. 
 
 Since these triangles share some common angles and 
 sides, they provide three equations with three unknown 
 values. These equations can be solved simultaneously, and 
 the lengths of the sides (or sensor-to-target distances) can 
 be determined. The position of each target relative to L3 
 can then be determined by projecting these lengths onto 
 the X and y axis. And, finally, the heading (h) of the tar- 
 gets relative to the centerhne of the cut can be determined 
 from two positions, as demonstrated in method 1. 
 
 Both of these methods determine the position and 
 heading of the targets on the continuous miner relative to 
 the MCS. Each target's position on the continuous miner 
 is known; therefore, the continuous miner's position and 
 heading relative to the desired cut can be determined. 
 
15 
 
 Figure 9.— Method 2 for determininq position and heading of 
 continuous miner relative to MCS. Angular position of three 
 targets (D,E,F) relative to one laser scanning device (L3) is 
 known. 
 
 Gyroscope 
 
 A mechanical gyroscope will be used to maintain an 
 accurate short-term knowledge of position and heading 
 from the most recent laser referenced location. Mechan- 
 ical gyroscopes have an inherent drift over time owing to 
 small mass imbalances in the rotor, and require periodic 
 recalibration to maintain accuracy. Therefore, the contin- 
 uous miner must reference its heading with respect to the 
 
 e^-eo 
 
 Sf-Se 
 
 Figure 10.— Mathematical representation of method 2 for posi- 
 tion determination. 
 
 desired cut from the laser-scanning system at regular 
 intervals. 
 
 When an accurate reading of heading is obtained from 
 the laser-scanning system, the gyroscope will be calibrated. 
 Any maneuvers involving turning desired by the continuous 
 miner can then be controlled and monitored with the 
 gyroscope. 
 
 The incorporation of a gyroscope will increase the reli- 
 ability of the laser referencing system by requiring the ref- 
 erence for the desired cut at regular intervals, rather than 
 continually. Therefore, the continuous miner will be able 
 to navigate for a short period of time when it cannot see 
 the reference or when the MCS referencing sensors fail or 
 are in error. At times when redundant, vaUd data are 
 obtained from these multiple sensory systems, it will be 
 combined to provide a greater and more reliable aware- 
 ness of the position and heading of the continuous miner. 
 
 A directional gyroscope, Model #DG57-0602-l devel- 
 oped by Humphrey, Inc., San Diego, CA, is to be em- 
 ployed on the continuous miner. It experiences a mechan- 
 ical drift of 0.1°/min. The desired accuracy in the heading 
 of a cut is approximately 1.4° (6 in per 20 ft). Therefore, 
 this gyroscope is usable as a heading indicator for 14 min 
 without recalibration, more than enough time for execution 
 of maneuvers requiring turning. 
 
16 
 
 Ultrasonic Ranging System 
 
 An ultrasonic sensor ranging system will be used to pro- 
 vide a knowledge of the physical surroundings of the face 
 area, i.e., the continuous miner's lateral position and head- 
 ing relative to the ribs for alignment, and obstacle detector 
 for forward and reverse tramming. In addition, the system 
 will serve as another source of relative heading of the 
 continuous miner. 
 
 A system developed by Denning Mobile Robots, Inc., 
 Woburn, MA, will be used. The system employs 24 
 addressable time-of-flight ranging units consisting of 
 environmental ultrasonic transducers developed for 
 operation in severe environments. Each ranging unit is 
 uniquely addressable and, upon request, returns a value 
 corresponding to the range from the specified unit to the 
 nearest object. 
 
 The ultrasonic ranging units will be mounted on the 
 continuous miner with their beams directed outward from 
 the machine. Lateral position and heading information 
 relative to the ribs will be determined from the units along 
 the length of the machine. It is not necessary that the 
 entire length of the continuous miner be covered in the 
 field-of-view of the ranging units for lateral position and 
 heading determination. However, the ranging units on the 
 front and rear of the machine should cover the entire 
 width of the continuous miner to provide reliable object 
 detection for collision avoidance when tramming. 
 
 Determining Lateral Position and Heading 
 Using Ultrasonic Ranging Units 
 
 The continuous miner will initially be assumed to be 
 positioned at the face and outby the MCS. Therefore, it 
 will rely on the ultrasonic ranging sensors to become posi- 
 tioned parallel and at a safe distance from the ribs for for- 
 ward tramming. 
 
 The information necessary to align the continuous 
 miner parallel to the ribs is the current angle of alignment 
 and its lateral position in the entry. This method of align- 
 ment is not as accurate as the laser-based scanning system, 
 owing to the wide beam width, the diffuse surface of mine 
 walls, and the potential presence of a line curtain. 
 
 The angle of alignment of the continuous miner with 
 respect to the ribs can be calculated from the measure- 
 ments of two lateral sensors. The measurements from all 
 the possible pairs of lateral sensors can be processed to 
 determine the most accurate alignment angle. When 
 monitored over time, they provide an indication of the 
 relative heading of the continuous miner. 
 
 Flux-Gate Heading Sensor 
 
 A flux-gate heading sensor, developed by KVH Indus- 
 tries, Middletown, RI, will be employed as a secondary 
 measurement of instantaneous changes in heading. The 
 absolute heading relative to true north will be unreliable 
 
 owing to magnetic interference from the continuous min- 
 ing machine. However, the sensor may provide accurate 
 instantaneous changes in heading. 
 
 A flux-gate heading sensor uses a solid-state electronic 
 sensor, which directly measures the Earth's magnetic field. 
 It is more reliable than a conventional magnetic compass, 
 since it does not use magnets or rotating cards delicately 
 balanced on jeweled bearings. It includes a micro- 
 processor-based system which takes thousands of readings 
 per second, interprets and averages them, and delivers the 
 resultant heading information. 
 
 The flux-gate heading sensor will be employed in the 
 same manner as the directional gyroscope. Tests will be 
 performed on the sensor to determine its precision and 
 reliability on the continuous mining machine. 
 
 Inclinometers 
 
 Accustar Electric Clinometers, pendulum-type incli- 
 nometers developed by Sperry Sensing Systems, Phoenix, 
 AZ, will be employed on the continuous miner to provide 
 the pitch-and-roU angle. These inclinometers employ a 
 fluid and gas-filled capacitive sensor whose capacitance 
 varies linearly with the rotation about its sensitive axis. 
 This variation in capacitcmce is transformed into an elec- 
 tronic signal and provides a direct measurement of tilt. 
 Two inclinometers with their sensitive axis at right angles 
 to each other will provide the pitch-and-roU angle of the 
 continuous miner. 
 
 Monitoring System 
 
 Data Acquisition with BCC52 Computer-Controller 
 
 The Micromint Basic-52 Computer-Controller (BCC52) 
 will gather the data from the previously mentioned sensors. 
 It is a stand-alone single board computer programmable in 
 BASIC or machine language. Three BCC52's will be em- 
 ployed, one dedicated to the laser-scanning system, one to 
 the ultrasonic ranging system, and one to the gyroscope, 
 compass, and inclinometers. 
 
 The function of the BCC52's is to gather the raw data 
 from their respective sensors, convert it into useful engi- 
 neering units, perform low-level sensor integration if nec- 
 essary, and deliver the data to a high-level computer. 
 
 Low-Level Sensor Fusion of Multiple 
 Laser-Based Scanning Systems 
 
 Multiple laser-scanning systems provide a redundant 
 source of information. Multiple values (V) of position and 
 heading are calculated from the laser-scanning systems. 
 By assigning weights (W) to the values and averaging this 
 information, a more accurate and rehable value for posi- 
 tion and heading of the continuous mining machine can be 
 determined. 
 
17 
 
 Each of the vcdues of position and heading [(V(1),V(2), 
 V(3)...V(n)] will be assigned a confidence level [(C(l), 
 C(2),C(3)...C(n)]. The confidence levels, representing the 
 accuracy and reliability of the laiser system, can be deter- 
 mined, in one way, from experiments of the system in a 
 controlled environment. The experiments indicate items 
 and conditions imder which the data are more accurate 
 £md rehable. For example, the data from the laser scanner 
 will most likely be more accurate when a greater angular 
 separation exists between the targets in its field-of-view or 
 when the target remains within a predetermined range. 
 Other indications of confidence level are the last known 
 position £md the current operating status of the machine. 
 This information could indicate that the confidence of one 
 laser scanner is low due to high dust concentrations, or 
 that the laser is operating outside its accurate range. 
 
 Average weights (W) for each value can be calculated 
 from the percentage confidence levels. 
 
 W(i) = C(i) / [C(l) + C(2) + ... + C(n)] 
 
 The fused value, V(f), of position jmd heading is then 
 determined by 
 
 V(f) = W(l) * V(l) + W(2) * V(2) + ... + W(n) * V(n) 
 
 These fused values for position and heading are deliv- 
 ered upon request to the integration system. 
 
 Integration System 
 
 The integration system will request the data from the 
 three monitoring systems, combine the position and head- 
 ing values, and create an appropriate model of the face 
 environment. 
 
 Integrating Position and Heading Information 
 
 The continuous miner's heading in the face area with 
 respect to the desired cut (MCS) can be determined from 
 the laser-scanning system, the gyroscope, and the ultra- 
 sonic ranging system. First, the laser system's reading of 
 heading, along with its associated confidence level, will be 
 requested from the laser monitoring system. Secondly, 
 the gyroscope's reading of heading change will be re- 
 quested. This heading change along with the most recent 
 referenced value of heading can be used to determine the 
 current heading. The confidence level of this value can be 
 obtained as a function of the length of time between read- 
 ings. The fmal value of heading will be requested from the 
 lateral sonar data. The associated confidence level can be 
 determined from the inherent inaccuracies of the ultra- 
 sonic rcinging units. 
 
 The values of heading of the continuous miner with 
 respect to the desired cut, along with their associated 
 confidence levels, can be used to determine a fused value 
 
 of heading and confidence. The method of averaging 
 weighted values, as discussed earUer, will be employed. 
 
 The value of lateral position of the continuous miner 
 with respect to the desired cut can be determined in a sim- 
 ilar manner, using the laser position data, and fusing it 
 with the value of lateral position change relative to the ribs 
 from the ultrasonic range data. The longitudinal position 
 of the continuous miner with respect to the desired cut can 
 only be determined by the laser system. Therefore, no 
 data fusion, other than the low-level fusion of multiple 
 laser units, is necessary for this information. 
 
 Modeling the Face Area 
 
 A geometric model indicating the continuous miner's 
 position and heading relative to the actual physical bound- 
 aries of the immediate area is essential for navigation. 
 The graphical technique of modeling the environment 
 developed by Turchan,* shows promise for the modeling 
 of a mine environment. This technique requires range 
 measurements from a designated center point on the vehi- 
 cle to the closest objects at incremental angles of view. 
 The data from the ultrasonic ranging system will provide 
 the information. 
 
 These range data, modeled as a function of angle, can 
 be mathematically analyzed to provide a knowledge of the 
 local environment. A large discontinuity in the function 
 represents the edge of an obstacle, or perhaps the pres- 
 ence of a crosscut. A discontinuity in the slope of the 
 function represents a point where two walls meet, at the 
 face or a corner. 
 
 Navigational Goal Scheduler 
 
 The navigational goal scheduler is a program, which will 
 function as the scheduler of the progress of the continuous 
 miner through the cutting sequence. It will contain two 
 methods of scheduling, sequential and contingency. The 
 schedulers observe and interpret the immediate world 
 model and generate a desired position and heading, or 
 goal state, of the continuous miner for the next sequence 
 of tramming actions. 
 
 The sequential goal scheduler has a predetermined list, 
 in memory, of goal states for the execution of an ideal cut. 
 If each of the goals is reached in the correct sequence, a 
 cut will have been performed. 
 
 The sequential scheduler analyzes the current position 
 and heading of the continuous miner relative to the cur- 
 rent goal state. If the goal was reached, the sequential 
 scheduler recommends a next goal state to the action plan- 
 ner. However, if the goal was not reached, control is 
 transferred to the contingency goal scheduler. 
 
 ^Turchan, M. P., and A. K. C. Wong. Low-Level Learning for a 
 Mobile Robot: Environmental Model Acquisition. Paper in Proceed- 
 ings of IEEE Conference on Robotics and Automation, St. Louis, MO, 
 Mar. 1985, pp. 156-161. 
 
18 
 
 The contingency goal scheduler is necessary when, for 
 one reason or another, the previous goal was not reached. 
 The apparent problem is determined and an alternative 
 goal state is recommended to the action planner. For 
 example, when the continuous miner is slipping or stuck, 
 the contingency goal scheduler will recognize the problem 
 and may recommend a goal state in the reverse direction 
 to try and free the machine. When the contingency goals 
 have been reached, control will be restored to the sequen- 
 tial goal scheduler. In addition, a history could be main- 
 tained containing contingency goals and the success of the 
 attempts at reaching these alternative goal states, to pro- 
 vide the continuous miner with the ability to learn from 
 its experiences. 
 
 Action Planner 
 
 The action planner is a program which plans, executes, 
 and monitors a sequence of machine maneuvers to enable 
 travel from the current position and heading to the goal 
 state designated by the navigational goal scheduler. This 
 program requires the knowledge of the primitive move- 
 ments and the basic maneuvers of the continuous mining 
 machine. 
 
 The primitive movements describe the speed and direc- 
 tion of tram for each track, and result in the basic maneu- 
 vers of the continuous mining machine. Each track can be 
 driven at two speeds, fast and slow, in both the forward or 
 reverse direction. 
 
 The basic maneuvers of the continuous miner are trans- 
 lation, pivot, and turn. A translational movement is pos- 
 sible in the forward or reverse direction, at fast or slow 
 speed. A pivot is a rotation about the center axis of the 
 continuous miner. It results when the two tracks are 
 driven in opposite directions, and is only possible at slow 
 speed. A turn is a rotation about one of the tracks, and 
 results when one track is driven at slow speed, and the 
 other track remains stationary. Because of the backlash in 
 track drives and track slippage, none of these movements 
 are performed with precision. 
 
 The current state, goal state, and the environmental 
 model are deUvered to the action planner from the 
 
 navigational goad scheduler. A sequence of primitive 
 movements is planned, by applying the rules of trigonom- 
 etry along with the knowledge of the mobiUty of the con- 
 tinuous miner. The sequence is executed by delivering the 
 maneuvers sequentially to the machine control computer. 
 The movement of the continuous miner is monitored 
 through request of machine response information from the 
 monitoring systems. 
 
 A time limit will be set on the action plaimer's attempts 
 to reach goal states. The action planner will continue to 
 plan, execute, and monitor machine maneuvers until this 
 time expires or the gocil state has been reached. If the 
 goad state is reached, control will be returned to the 
 sequential goal scheduler. However, if the time limit has 
 expired, a navigational difficulty is occurring, such as 
 the machine being stuck. In that case, control will be 
 transferred to the contingency goal scheduler for a recom- 
 mendation of an alternative goal state. 
 
 Navigational Test-Beds 
 
 Navigational experiments to evaluate sensor perform- 
 ance and guidance algorithms will be conducted on two 
 mobile test-beds. The Joy 16CM miner-bolter is the avail- 
 able continuous mining machine for testing of navigational 
 concepts (fig. 11). It is a drum-type continuous mining 
 machine designed to extract coal in an underground mine 
 using the technique of room-and-pillar mining. Another 
 test-bed to be used for research into autonomous naviga- 
 tion in mines is a locomotion emulator being developed 
 for the Bureau by Carnegie Mellon University, Pittsburgh, 
 PA. This test-bed will have the ability to emulate the loco- 
 motion of many different mining vehicles currently in use, 
 and is extendible to any future type of vehicles. 
 
 Initial attempts at autonomous navigation will use the 
 locomotion emulator as a full-scale model of a smaller 
 continuous mining machine (e.g., Joy 12CM). The loco- 
 motion emulator is a much more manageable mechanism 
 to test the basic theories and concepts of navigation in 
 mines. The Joy 16CM will initially be used to test the 
 reHability of various navigational sensors on an actual 
 mining machine. 
 
19 
 
 ^, 
 
 / 
 
 ^ 
 
 
 a 
 
 ^ ^ 
 
 
 r — ' 
 — . / 
 
 ---• '.A 'T" 1 
 
 \) 1 
 1 l' 
 
 1 ll 
 
 1 ' 
 
 I 1 1 
 
 
 > . 
 
 1 
 
 1 
 
 
 /r-.-.--i.---z\^p^ 
 
 Y 
 
 16' 
 
 
 
 
 1 
 
 1 ■ " 
 
 1 
 
 1 
 
 
 
 
 ■ — U.'-> 1 
 
 Figure 11.— Joy 16CM available continuous mining machine for testing of navigational concepts. 
 
20 
 
 CONCLUSIONS 
 
 The Bureau is proposing to develop a navigational guid- 
 ance system to provide a continuous mining machine with 
 the ability to navigate autonomously through the positions 
 and headings necessary to execute straight-cuts and cross- 
 cuts in room-and-pillar type mining. These cuts must also 
 be executed according to the mine plan in order to main- 
 tain a safe and productive mine. This type of navigation 
 has been classified as face navigation. Although two other 
 types of navigation in mines are known to be needed, i.e., 
 local and global, autonomous face navigation has been 
 chosen as the short-term goal. Successful navigation at the 
 face will result in a great increase in the overall safety of 
 the mine worker, and many of the research issues of face 
 navigation are applicable to the other types of navigation. 
 
 Autonomous face navigation requires that the contin- 
 uous miner be provided with an indication of the position 
 and heading of the desired cut, a knowledge of its physical 
 surroundings, and precise guidance abilities to control and 
 monitor its movements at the face. 
 
 A MCS will be employed as a reference for the desired 
 cut in the face area. The MCS will be precisely positioned 
 in the face area to provide a fixed, stable reference for 
 both the position and heading of the desired cut. The 
 continuous miner will refer to the MCS for alignment and 
 guidance during the cutting process. 
 
 A sensory configuration has been outlined for the navi- 
 gational guidance of the continuous miner. Multiple sen- 
 sor systems have been chosen to provide a more accurate 
 and reliable awareness of the environment. Optical laser- 
 based scanning devices will provide the continuous miner 
 with the ability to reference the MCS. A mechanical 
 directional gyroscope will be mounted on the continuous 
 
 mining machine to provide relative heading information. 
 A flux-gate heading sensor will be employed in a similar 
 manner, and tests will be performed to determine its reli- 
 abiUty with the magnetic interference present on the con- 
 tinuous mining machine. Ultrasonic ranging units will be 
 used to provide lateral position and heading information 
 for alignment, and a front and rear obstacle detector for 
 treunming of the continuous miner. Finally, two pendulum- 
 type incUnometers will be employed on the continuous 
 miner to provide the pitch-and-roU angle. 
 
 Three monitoring systems consisting of the BCC52 
 computer-controller will be used for acquisition of the 
 sensory data. The BCC52's will gather the raw data from 
 the sensors, convert it to useful engineering units, perform 
 low-level sensor fusion if necessary, and deliver the data to 
 the integration system. 
 
 An integration system consisting of a BCC52 computer- 
 controller will request the data from all the monitoring 
 systems, integrate the position and heading of the contin- 
 uous mining machine with respect to the desired cut, and 
 create an appropriate model of the face environment. 
 
 The navigational goal scheduler is a program which will 
 schedule the progress of the continuous miner through the 
 cutting sequence. This program will observe and interpret 
 the model of the face area, and generate the next desired 
 position and heading of the continuous miner. The current 
 position and heading, and the next desired position and 
 heading will be delivered to an action planning program. 
 The action planner will generate, execute, and monitor a 
 sequence of tramming actions to reach the next desired 
 position and heading. 
 
21 
 
 APPENDIX.-RESEARCH ISSUES IN AUTONOMOUS NAVIGATION 
 
 An autonomous continuous mining machine requires 
 methods of acquiring, processing, and manipulating 
 navigational information to interact with objects in the 
 mining environment. The methods used by autonomous 
 vehicles in other environments are presented along with 
 their apphcabUity for autonomous navigation of a 
 continuous mining machine in a mining environment. 
 
 METHODS OF ACQUIRING NAVIGATIONAL 
 INFORMATION 
 
 The fundcimental requirement for autonomous vehicle 
 navigation in any environment is to acquire a represen- 
 tation of the physical surroundings. This representation 
 consists of the position of objects relative to the vehicle, 
 and can be obtained from sensors providing range mea- 
 surements from the vehicle to the object. 
 
 Ranging Sensors 
 
 The continuous miner ranging requirements can be clas- 
 sified into two areas: positioning and collision avoidance. 
 Positioning sensors require high accuracy and precision 
 over a wide range, with a nairrow field-of-view. Collision 
 avoidance sensors do not require the high accuracy, preci- 
 sion, and range of positioning sensors. However, they do 
 require a large field-of-view, and the outer range of the 
 sensor should provide the continuous miner the time to 
 stop when a collision is imminent. 
 
 The following is a description of some of the methods 
 of rcinging, which are employed for both positioning and 
 collision avoidance on existing autonomous vehicles. The 
 various techniques for ranging are discussed, followed by 
 the sensing mediums employing these technologies, with 
 attention focused on their functionaHty on a continuous 
 mining machine in the mine. 
 
 Techniques for Ranging 
 
 Time-of-Flight 
 
 The time-of-flight ranging technique provides range to 
 an object by measuring the time it takes for a pulse of 
 energy to travel from a transmitter to the object and back 
 to a receiver. The range can then be calculated by multi- 
 plying the velocity of the pulse by one-half of the time 
 required to travel the distance. 
 
 The time-of-flight sensors maintain reliability as long as 
 the return signal is received and detected. The reception 
 and detection of the return signal is dependent on the dis- 
 tance to the object, the strength of the transmitted signal, 
 and the characteristics of the reflecting surface. As the 
 range to the object increases, the intensity of the return 
 
 signal decreases. Obviously, a transmitted signal of greater 
 strength provides a return signal at greater ranges. How- 
 ever, a maximum range exists where a return signal can no 
 longer be detected by the receiver. No signal would be 
 received when the tremsmitted signal is reflected from a 
 smooth object whose surface is at an angle from the trans- 
 mitter. However, these specular reflections will not 
 present many difficulties in a mine environment, where 
 the diffuse surface of the mine walls are most often the 
 reflecting object for range measurements. It is likely, 
 however, that the transmitted signal will scatter after en- 
 countering the rough surface, and those signals could re- 
 flect from secondary objects and provide a false return 
 signal to the receiver. 
 
 The appUcations of time-of-flight ranging systems are 
 related to the beam width of the signal being transmitted. 
 The distance determined by time-of-flight ranging systems 
 is the distance to the closest point which the signal encoun- 
 ters. If the transmitted signal has a wide beam width, the 
 exact angular position of the object relative to the sensor 
 is not known. Wide beam, time-of-flight ranging systems 
 cannot be used for precise object positioning, but are very 
 appropriate for collision avoidance. 
 
 Triangulation 
 
 The triangulation technique of ranging requires two 
 sensors at a known separation on the vehicle providing the 
 angle to the same object. It utilizes the basic trigonom- 
 etric property of triangles. That is, when given the length 
 of one side (separation between the sensors) and two an- 
 gles of a triangle (angles from the sensors to the object), 
 the lengths of the two remaining sides (ranges from the 
 sensors to the object) of the triangle can be calculated. 
 
 The range accuracy of a system employing the tech- 
 nique of triangulation is dependent on the accuracy of the 
 measured angles and, furthermore, the variations in angu- 
 lar position to an object decrease as a function of distance. 
 
 One type of triangulation ranging, similar to human 
 vision, uses two camera images. The range to the object 
 is inversely proportional to the displacement of the object 
 on the image planes. The ranging becomes more accurate 
 the further apart the imaging devices are, or the closer the 
 object is. However, if the imaging devices are too far 
 apart or the object is too close, the object may not appear 
 in both images. 
 
 The major difficulty of this type of ranging system is 
 that the same object has to be identified accurately in both 
 images. These systems typically use the ambient light to 
 illuminate the object. In a mining environment, however, 
 an artificial source of light must be provided, and not 
 many objects are present in a mine which are easy to 
 identify. 
 
22 
 
 Another method of triangulation ranging is demon- 
 strated in the commercially available Lasernet system, 
 which can measure and report the angular location of 
 special cylindrical retro-reflective targets. The system 
 consists of a scanning laser source and a photodetector. 
 When the laser hits the retro-reflective target, a reflected 
 beam is returned and detected by the photodetector. Since 
 the laser is scanned at a constant angular velocity, the time 
 between the steu^t of the laser's sweep and the moment at 
 which the reflected beam returns from the target can be 
 used to calculate angular position. The angular positions 
 of the target relative to two Lasernet units and the sepa- 
 ration between these units, can be used to determine the 
 range to the target. 
 
 In summary, triangulation ranging systems are limited 
 in range accuracy by their ability to measure small angles 
 precisely. The long-range accuracy can be improved by 
 increasing the separation between the two sensors. 
 
 Interferometry 
 
 Interferometry offers extremely precise and accurate 
 distance measurements. Interferometry is based on the 
 interference patterns that result between two energy waves 
 that travel paths of different lengths. If the length of the 
 path for one of the energy waves is increased, the two 
 waves interact and produce constructive and destructive 
 interference. By observing this interference pattern and 
 knowing the wavelength of the source, it is possible to 
 calculate the relative distance the two energy waves have 
 traveled at fractional wavelength accuracies. This distance 
 can be the change in position of a target on a moving 
 vehicle relative to some previous position. 
 
 A retro-reflector target is required on the vehicle in 
 order to provide a reliable return signal for the interfer- 
 ometer. The distance measured by interferometry ranging 
 is the change in distance along the straight-line path from 
 the transmitter to the target. The angles of this straight- 
 line path must be known to determine the displacement of 
 the target in each coordinate direction. 
 
 The disadvantage of interferometry ranging is that it 
 provides only relative distance measurement. Therefore, 
 the usable measurements are cumulative and require a 
 constant line-of-sight between the target and system. 
 Interferometry would not be reliable in a mining environ- 
 ment owing to the frequent occurrence of loss of the line- 
 of-sight caused by other machinery, dust, or rocks. 
 
 Sensing Mediums for Ranging . 
 
 Acoustic Signals 
 
 Acoustic ranging can be accomplished by using the 
 previously discussed techniques of time-of-flight and tri- 
 angulation. Ultrasonic energy (acoustic energy of fre- 
 quencies above the limits of human hearing) is commonly 
 used for ranging devices on mobile vehicles. 
 
 The wide beam width of ultrasonic waves introduces 
 some uncertainty in the perceived distance to and the 
 anguleir position of an object. The range obtained is the 
 distance to the first object in the field-of-view of the sen- 
 sor, and may not be the distance to the object in the beam 
 centerline. Therefore, it functions best as a collision 
 avoidance sensor, by detecting the presence of the nearest 
 object in a wide field-of-view. 
 
 An ultrasonic wave may not be reflected back to the 
 receiver if it encounters a specular surface at an angle. 
 Therefore, ultrasonic ranging devices function much more 
 reliably in an environment with diffuse surfaces, such as in 
 a mine. These surfaces scatter the transmission and most 
 often reflect a signal back to the receiver. The use of 
 acoustic sensors for navigation in mines was demonstrated 
 with Carnegie Mellon University's Terregator vehicle.^ 
 The vehicle was tested at the Bureau experimental mine, 
 and successfully navigated in the mine environment using 
 ultrasonic ranging sensors. 
 
 Optical Ranging 
 
 Optical ranging can be accomplished by using the pre- 
 viously discussed techniques of time-of-flight, triangulation, 
 and interferometry. Imaging systems employing vision are 
 commonly used for guidance or road following, and optical 
 laser-based ranging systems are commonly used for precise 
 positioning of autonomous vehicles. 
 
 Any type of ranging system employing vision imaging 
 devices are not presently being considered for use in the 
 mine. There is no natural source of ambient light, thus 
 artificial illumination would have to be provided. Also, the 
 amount of data manipulation necessary is too great to jus- 
 tify vision over simple ranging devices. 
 
 Optical ranging devices employing laser sources provide 
 a quick and accurate method of ranging. They have a very 
 narrow beam width and are most appropriate for the pre- 
 cise positioning information required during the execution 
 of the cutting sequence at the face. 
 
 Radar 
 
 Radar uses the time-of-flight technique to determine 
 the range to an object. Radar employs radio frequency 
 energy. It has an advantage over optical and acoustical 
 waves in that it can propagate through dust. 
 
 Radar systems employing electromagnetic waves are 
 best suited to very long ranges. The high velocity of elec- 
 tromagnetic waves induces the possibility of transmitter- 
 receiver interference at shorter ranges. Therefore, expen- 
 sive and complicated circuitry is necessary to produce a 
 signal that is short in duration. 
 
 'Oppenheim, I., and W. Whittaker. Demonstration of Robotic Map- 
 ping of Mine Spaces. Eng. Const. Rob. Lab., Camegie-Mellon Univ., 
 Pittsburgh, PA, Apr. 1985, 58 pp. 
 
23 
 
 Microwave energy waves are also used in radar systems. 
 Ranging devices employing these waves can be configured 
 to function over ranges suitable to mobile vehicle re- 
 quirements, and provide collision avoidance by detecting 
 the presence (not precise position) of an object. They are 
 best suited as presence detectors, such as level indicators 
 or back-up alcu^ms for manned vehicles. However, they 
 provide Uttle resolution for autonomous vehicle 
 apphcations. 
 
 METHODS OF PROCESSING NAVIGATIONAL 
 INFORMATION 
 
 Mapping Techniques 
 
 The continuous miner requires a geometric model of 
 the local environment. One possible method of environ- 
 mental modeling of the local area in a mine is the graph 
 synthesis approach, developed by M. P. Turchan.^ This 
 approach requires range measurements to the closest 
 objects around the vehicle at incremental angles of view. 
 
 A center point on the machine is chosen as an origin 
 point for the range measurements. The model is obtained 
 by examining the value of range from the origin to the 
 nearest object as a function of the angle-of-view of the 
 sensor (0 to 360°). 
 
 Applications of Graph Synthesis Model 
 to Mine Model 
 
 The range data from the continuous miner's sensors 
 modeled as a function of angle can aid in path planning 
 and entry navigation. The discontinuities of the function 
 itself represent the edge of an obstacle, or perhaps the 
 
 ^urchan, M. P., and A. K. C. Wong. Low-Level Learning for a 
 Mobile Robot: Environmental Model Acquisition. Paper in Proceed- 
 ings of IEEE Conference on Robotics and Automation, St. Louis, MO, 
 Mar. 1985, pp. 156-16L 
 
 presence of a crosscut. The discontinuities of the slope of 
 the function represent the points where two walls meet, 
 possibly at the face or at a corner. These representations 
 of the environment can be an aid to both local and global 
 navigation. 
 
 INTELLIGENT TECHNIQUES FOR 
 NAVIGATIONAL CONTROL 
 
 A method of global path planning is necessary for the 
 continuous mining machine. This planning requires a map 
 programmed in memory indicating the paths of travel in a 
 mine. 
 
 Global Path Planning 
 
 The continuous miner must be able to determine a safe 
 and traversable path to a destination given a reliable map 
 of the area between the current position and destination. 
 Many popular techniques exist for finding the optimal path 
 and are used by other autonomous vehicles. 
 
 Many techniques of path planning for other mobile ro- 
 bots involve depicting the maps in a graph form with costs 
 assigned to possible routes. In the case of mine travel, 
 those costs may represent distance, risk, maneuverability, 
 or traversabillty. A search Is performed on the graph, and 
 the lowest cost path is chosen. 
 
 However, global path planning in a mine does not re- 
 quire complex programming algorithms. Most mines have 
 designated main passageways for travel in a particular 
 direction. Since most of the travel is done along these 
 routes, the only planning necessary is for the short travel 
 from the main passageways to the work area. The plan- 
 ning techniques are uncomplicated due to the regular and 
 parallel nature of paths in a mine. This planning involves 
 choosing a path from among many similar parallel entries 
 and crosscuts. Any safe path with a minimal number of 
 turns can be chosen. A few simple heuristics can provide 
 the global path planning necessary in a mine. 
 
 U.S. GOVERNMENT PRINTING OFFICE: 611-012/00,053 
 
 INT.BU.OF MINES,PGH.,PA. 28881 
 
"0 
 
 m 
 
 z 
 
 33 T) O -D 03 C 
 
 
 P 
 
 
 
 
 3" 
 
 5 
 Q. 
 
 1 ^ 
 
 CD • 
 
 ^o 
 
 O" 
 
 c 
 3 
 
 
 
 Crt 
 
 c 
 
 
 ^ CD 
 
 55 
 
 (Q 
 
 X 
 
 -A. 
 
 o' 
 
 13 
 
 -D 
 
 ICIAL 1 
 FOR PR 
 
 > 
 
 CD 
 
 
 
 
 i; 
 
 
 
 BO 
 D 
 
 — » 
 
 rtment c 
 Mines 
 
 > S 
 
 t>0 
 CO 
 
 
 
 
 Q. 
 
 c 
 
 — ». 
 5 
 
 H 0) 
 
 
 
 
 0" 
 
 CD 
 
 ?z 
 
 
 
 
 3 
 
 5" 
 
 ffiS3 
 
 
 
 
 
 CD 
 
 I^ 
 
 
 
 
 
 o' 
 
 ts 
 
 
 
 
 
 
 o 
 
 
 
 
 
 
 o 
 
 
 
 
 
 
 m 
 O 
 
 c 
 > 
 
 o 
 
 "D 
 13 
 O 
 33 
 
 m 
 
 r— 
 O 
 -< 
 m 
 3 
 
 C 214 8 9 
 

 
 
 
 -Ho^ 
 
 "^ A^ **" 
 
 
 
 
 
 
 
 0" ♦' 
 
 o «J 
 
 
 
 
 
 
 «-i°*. 
 
 
 ^0 
 
 
 
 
 
 
 
 ,V 
 
 
 
 A^ 
 
 
 
 o . .* -A. 
 
 
 
 • <^^ 
 
 f.* r.^" 
 

 
 
 -^«^^ 
 
 r. o^^ ^^^ •• 
 
 4 o 
 
 I>*.f' «^ aP 
 
 ^o. 
 
 •Jc«5^^tafc*- o 
 
 
 5- o. 
 
 
 
 
 
 
 "-o^<b'^ 
 
 
 :. '-n^o^ /, 
 
 
 V-* e « .-?) 
 ^* > V 
 
 ^ • '^J. AV ♦ 
 
 ^* <^ \ ' ,«^ 
 
 
 >> ..V.., V 
 
 
 
 
 
 
 c-^.l^'>o ^/.y^/"-^. o°^l^^^'\ /' 
 
 
 ''^^.<:,^'^ 
 
 vP 'j 
 
 J"^^ 
 
 V ^ 
 
 
 HECKMAN 
 BINDERY INC. 
 
 #AUG 89 
 N. MANCHESTER, 
 INDIANA 46962 
 
 
 5-^ w-.