APF-BDCA: Improved A-Star Path Planning Algorithm for Intelligent Vehicles Integrating Artificial Potential Fields and Direction Constraint

Abstract

Path planning is a key research focus in the control technology of unmanned systems, such as intelligent vehicles and robots. The conventional A* algorithm suffers from low search efficiency, redundant paths, and impractical solutions; also, the planned path is often too close to obstacles, and the vehicle behavior does not meet realistic dynamics when turning. All of these factors prevent it from being effectively used in practical applications such as the autonomous navigation of intelligent vehicles and robots. In this paper, an improved A* algorithm is proposed. First, the artificial potential field method is introduced to guide the path nodes away from the different level dangerous areas near obstacles to increase the safety of the path. Second, due to the restriction of the vehicle driving direction by lane, a strategy is proposed to constrain the expansion direction of nodes, and the unidirectional <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$\mathrm{A}^{\star}$</tex> algorithm is extended to the bidirectional A* algorithm, which improves computational efficiency while meeting the requirements for the forward direction in specific road sections. Finally, redundant point processing strategies and cubic B-spline curves are used to further optimize and smooth the path. Experimental results show that the performance of the improved algorithm is superior to the original algorithm. It shortens the path length, reduces the runtime of the algorithm, eliminates path points that are too close to the edges and sharp curves, and resolves the issue of vehicles driving against the traffic on certain roads, thus achieving significant optimization.