Abstraction-based Planning for Uncertainty-aware Legged Navigation

This paper addresses the problem of temporal-logic-based planning for bipedal robots in uncertain environments. We first propose an Interval Markov Decision Process abstraction locomotion (IMDP-BL). Motion perturbations from multiple sources uncertainty are incorporated into our model using stacked Gaussian process learning order to achieve formal guarantees on behavior system. consider tasks which can be specified Linear Temporal Logic (LTL). Through a product IMDP construction combining IMDP-BL robot and Deterministic Rabin Automaton (DRA) specifications, we synthesize control policies allow safely traverse environment, iteratively unknown dynamics until specifications satisfied with satisfactory probability. demonstrate methods simulation case studies.