Answers are grounded in this paper and cite their sources.
Efficient trajectory planning in dynamic urban environments requires fast discrete decision making while preserving the feasibility and continuity of vehicle motion. This paper develops a path–velocity decoupled hierarchical planner in which both path and speed planning are organized into behavior and motion levels. The main algorithmic contribution lies in a bidirectional alternating A* (BirA*) path-search strategy that combines a 32-direction neighborhood with angle-based pruning, hybrid and parent-node heuristic evaluation, turning cost, and dynamically adjusted search weights to reduce unnecessary search effort. The discrete behavior decisions are subsequently mapped into convex feasible regions and refined by quadratic programming, while the temporal interactions associated with moving obstacles are handled downstream in the space–time domain through longitudinal behavior planning and constrained speed optimization. Controlled ablation experiments show that the full BirA* configuration reduces search time, expanded nodes, and generated candidates by approximately 55.9%, 61.6%, and 63.7%, respectively, relative to a controlled 32-direction fixed-Manhattan baseline. Simulations covering static-obstacle avoidance, overtaking, and unprotected left turns, together with seven predeclared obstacle-interaction cases, further evaluate the planner under structured dynamic-road conditions.
No takes yet. Share an insight, caveat, or question.
Li et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: