We have seen an unprecedented expansion of technology in our daily lives in past decades. However, we still far from idealized future when intelligent systems are part of our quotidian life. As they directly operate in the physical world, they must achieve high-level task specifications with performance guarantees. Much recent progress towards this goal has been made through an automatic controller synthesis from temporal logic specifications. Existing approaches, however, have been limited to relatively short and simple specifications. Furthermore, existing methods generally require conservative assumptions and do not scale well with system dimensionality. To promote long-term autonomy for these intelligent systems, our research focus on two main aspects: taskable and adaptive.Taskable intelligent systems can interpret and plan tasks from high-level instructions and execute actions which can adapt to a particular context in which the system is operating. Hence, we propose a scalable, provably complete algorithm that synthesizes continuous trajectories to satisfy non-convex temporal logic specifications. We separate discrete task planning and continuous motion planning on-the-fly and harness highly efficient Boolean satisfiability (SAT) and trajectory optimization solvers to find dynamically feasible trajectories that satisfy non-convex specifications for high-dimensional systems. The proposed design algorithms are proven sound and complete, and simulation results demonstrate our approach's scalability. Adaptive intelligent systems are aware of their capabilities and limitations to overcome uncertainties and learn from their own experiences from observation, planning accordingly. Thus, we propose a framework for scalable task and motion planning in uncertain environments that combines the best of belief-space planning and symbolic control. Specifically, we provide a counterexample-guided-inductive-synthesis algorithm for the robust satisfaction of probabilistic Signal Temporal Logic (PrSTL) specifications in the belief space. Our method automatically generates actions that improve confidence in a belief state when necessary, thus using active perception to satisfy PrSTL specifications.