A Two-Stage Optimization-Based Motion Planner for Safe Urban Driving

Recent road trials have shown that guaranteeing the safety of driving decisions is essential for wider adoption autonomous vehicle technology. One promising direction to pose requirements as planning constraints in nonlinear, nonconvex optimization problems motion synthesis. However, many implementations this approach are limited by uncertain convergence and local optimality solutions achieved, affecting overall robustness. To improve upon these issues, we propose a novel two-stage framework: In first stage, find solution mixed-integer linear programming (MILP) formulation synthesis problem, output which initializes second nonlinear (NLP) stage. The MILP stage enforces hard rule compliance generating right subspace, while NLP refines within bounds feasibility smoothness. We demonstrate effectiveness our framework via simulated experiments complex urban scenarios, outperforming state-of-the-art baseline metrics convergence, comfort, progress.