Abstract
The Rapidly-exploring Random Tree Star (RRT*) algorithm, widely utilized for path planning, faces challenges, such as slow acquisition of feasible paths and high path costs. To address this issue, this paper presents an improved algorithm based on RRT* that can obtain high-quality paths faster, termed Faster High-Quality RRT*(FHQ-RRT*). The proposed algorithm enhances the exploration efficiency and path quality of mobile robots through three key innovations: First, a dynamic sparse sampling strategy that adaptively adjusts the sampling density according to the growth rate of the random tree, thereby increasing the algorithm's growth speed while maintaining adaptability to complex environments. Second, a new node creation method that combines the bisection method, triangle inequality, and the concept of KeyPoints to reduce the cost of creating new nodes. Third, a focused rewiring strategy that restricts the rewiring operation to valuable regions, thereby improving rewiring efficiency. The performance of FHQ-RRT* was validated in four simulation maps and compared with other algorithms. In all validated maps, FHQ-RRT* consistently achieved the lowest path cost. Regarding time cost, FHQ-RRT* reduced the planning time by over 40% in the circular-obstacle map, 77% in the simple maze map, 56% in the complex maze map, and 50% in the narrow map. The simulation results show that FHQ-RRT* can rapidly generate high-quality paths faster than other algorithms.