MPTP: Motion-planning-aware task planning for navigation in belief space

We present an integrated Task-Motion Planning (TMP) framework for navigation in large-scale environments. Of late, TMP manipulation has attracted significant interest resulting a proliferation of different approaches. In contrast, received considerably less attention. Autonomous robots operating real-world complex scenarios require planning the discrete (task) space and continuous (motion) space. knowledge-intensive domains, on one hand, robot to reason at highest-level, example, objects procure, regions navigate order acquire them; other feasibility respective tasks have be checked execution level. This presents need motion-planning-aware task planners. this paper, we discuss probabilistically complete approach that leverages task-motion interaction navigating large returning plan is optimal task-level. The intended motion under sensing uncertainty, which formally known as belief planning. underlying methodology validated simulation, office environment its scalability tested larger Willow Garage world. A reasonable comparison with work closest our also provided. demonstrate adaptability by considering building floor domain. Finally, limitations put forward suggestions improvements future work.