Optimal Probabilistic Motion Planning With Potential Infeasible LTL Constraints

This paper studies optimal motion planning subject to and environment uncertainties. By modeling the system as a probabilistic labeled Markov decision process (PL-MDP), control objective is synthesize finite-memory policy, under which agent satisfies complex high-level tasks expressed linear temporal logic (LTL) with desired satisfaction probability. In particular, cost optimization of trajectory that infinite horizon considered, trade-off between reducing expected mean maximizing probability task analyzed. The LTL formulas are converted limit-deterministic Büchi automata (LDBA) reachability acceptance condition compact graph structure. novelty this work lies in considering cases where specifications can be potentially infeasible developing relaxed product MDP PL- LDBA. allows revise its plan whenever not fully feasible quantify revised plan’s violation measurement. A multi- problem then formulated jointly consider satisfaction, respect original constraints, implementation policy execution. solved via coupled programs. first bridges gap revision potential synthesis both prefix suffix over horizons. Experimental results provided demonstrate effectiveness proposed framework.