Autonomous vehicle (AV) path planning is a core decision-making function that ensures safe, efficient, and comfortable navigation in dynamic traffic environments. This review synthesizes recent advances in AI-driven path planning by systematically categorizing algorithms into graph-based, sampling-based, optimization-based, learning-based, and hybrid frameworks. Key performance objectives, including collision avoidance, ride comfort, energy efficiency, and real-time feasibility are discussed alongside representative methods such as A , D*, RRT/RRT *, model predictive control, heuristic/metaheuristic optimizers, deep reinforcement learning, and imitation learning. The review further analyzes cooperative path planning enabled by V2X connectivity, highlighting coordination strategies for platooning, merging, and intersection negotiation. Challenges including uncertainty handling, generalization across scenarios, computational constraints, safety assurance, and sim-to-real transfer are critically examined. Finally, open research directions are presented, emphasizing unified evaluation benchmarks, interpretable learning-based planners, multi-agent safety guarantees, and robust planning under imperfect perception. The paper provides an updated reference framework for researchers and practitioners seeking scalable and deployable path planning solutions for next-generation AV systems.
Baskar Ponnusamy (Fri,) studied this question.