Hybrid Path Planning for Robotics: APF-RRT Enhancement with Sobol Sequence Integration

The current study presents a novel path-planning algorithm, APF-IRRT*-SB, which integrates the Artificial Potential Field (APF) method with an enhanced Rapidly-Exploring Random Tree Star (RRT*) algorithm using the SobolBurkardt sequence for improved performance in dynamic and complex environments.The proposed algorithm addresses key limitations of traditional methods, such as computational inefficiency and suboptimal path generation, by incorporating a probabilistic goal-basing mechanism and low-randomness sampling for smoother and shorter paths.Comparative experiments were conducted on two scenarios: mass point motion and a two-link robot arm, tested across three environments of varying complexity.The experiments, performed using MATLAB R2023a on a Core i5-12450H CPU with 8GB RAM, demonstrated that APF-IRRT*-SB reduced path length by approximately 4% and computation time by 50% compared to APF-RRT* in the mass point case.For the two-link robot arm, the proposed method improved path length by up to 69.5% and reduced iterations by 94.4% compared to RRT*, while showing a 5.6% improvement in path length and an 82.5% reduction in iterations over APF-RRT*.These findings highlight the proposed algorithm's potential for efficient path planning in robotics, particularly in industrial and logistics applications requiring adaptability and computational speed.