Optimized PAB-RRT Algorithm for Autonomous Vehicle Path Planning in Complex Scenarios

Path planning is a pivotal technology for autonomous vehicles, directly influencing driving safety and comfort. Developing algorithms adaptable to diverse scenarios is critical for ensuring the safe operation of autonomous driving systems and advancing their engineering applications. The existing Rapidly exploring Random Tree (RRT) algorithm has limitations such as low efficiency and tortuous, lengthy paths. To address these issues, this study proposes the PAB-RRT algorithm, which integrates probabilistic goal bias, adaptive step size, and bidirectional exploration into RRT. Comparative simulations were conducted to evaluate PAB-RRT against traditional RRT, RRT*, and single-strategy improved variants (A-RRT, P-RRT, B-RRT). Results show that in static multi-obstacle scenarios, PAB-RRT completes planning with 30 iterations (6.99% of traditional RRT), 0.1255 s computation time (21.9% of traditional RRT), and a 130.83 m path length (7.2% shorter than traditional RRT). In dynamic obstacle scenarios, it requires 19 iterations (0.0434 s) at the initial stage and 37 iterations (0.0861 s) after obstacle movement, with path length stably around 130 m. Overall, PAB-RRT outperforms traditional algorithms in exploration efficiency, path performance, and robustness in complex settings, better meeting the efficiency and reliability requirements of autonomous vehicle path planning under complex scenarios and providing a feasible reference for related technology.