Receding Horizon Control-Based Motion Planning With Partially Infeasible LTL Constraints

This work considers online optimal motion planning of an autonomous agent subject to linear temporal logic (LTL) constraints. The environment is dynamic in the sense containing mobile obstacles and time-varying areas interest (i.e., reward workspace properties) be visited by agent. Since user-specified tasks may not fully realized partially infeasible), this hard soft LTL constraints, where constraints enforce safety requirement (e.g. avoid obstacles) while represent that can relaxed strictly follow user specifications. generate policies, decreasing order priority, 1) formally guarantee satisfaction constraints; 2) mostly satisfy minimize violation cost if desired are infeasible); 3) optimize objective rewards collection visiting more interests). To achieve these objectives, a product automaton, which allows constructed. A utility function developed quantify differences between revised plan, accumulated designed bias plan towards those interests. Receding horizon control synthesized with formula maximize utilities over finite horizon, ensuring satisfied satisfied. Simulation experiment results provided demonstrate effectiveness strategy.