A Vehicle Path Planning Algorithm: QDDG-RRT

Abstract

Autonomous vehicles require highly reliable collision-free capabilities, necessitating extensive research in path planning. Path planning determines an optimal path, crucial for safe and efficient driving. The Rapidly-exploring Random Tree (RRT) algorithm, while widely used, suffers from slow search speeds, numerous inflection points, and redundant operations. To address these issues, We propose a global path planning algorithm: Quick Dynamic Directional Guidance-RRT (QDDG-RRT) algorithm. Key improvements include dynamically constraining the search space using a direction guidance strategy, employing steering techniques to avoid obstacles, optimizing path length to minimize global cost, and refining trajectories using second-order Bessel curves. Simulation experiments compare QDDG-RRT with P-RRT(Probabilistic Rapidly-exploring Random Tree), P-RRT*, APF(Artificial Potential Field), and A* algorithms. Results show that QDDG-RRT outperforms others in execution speed, path length, and smoothness, effectively avoiding obstacles and maintaining safe distances in complex environments.

Keywords

• RRT
• direction guide
• directional rotation
• path cost minimization
• Bessel curve