Automatic Trajectory Synthesis for Real-Time Temporal Logic

Many safety-critical systems, such as autonomous vehicles and service robots, must achieve high-level task specifications with performance guarantees. Much recent progress toward this goal has been made through an automatic controller synthesis from temporal logic specifications. Existing approaches, however, have limited to relatively short simple Furthermore, existing methods either consider some prior discretization of the state space, deal only a convex fragment logic, or are not provably complete. We propose scalable, complete algorithm that synthesizes continuous trajectories satisfy nonconvex over reals (RTL) separate discrete planning motion on-the-fly harness highly efficient Boolean satisfiability linear programming solvers find dynamically feasible RTL for high-dimensional systems. The proposed design algorithms proven sound complete, simulation results demonstrate our approach's scalability.