The Improved RRT Integrated With the Artificial Potential Field Path Planning Algorithm

摘要

The traditional informed rapidly exploring random tree * algorithm (IRRT*) has several drawbacks, including low efficiency, numerous ineffective samples, strict requirements of the environment, and slow convergence of path length in narrow passages. This study proposes an improved IRRT* integrated with an artificial potential field path planning algorithm (IRRT*-APF) by introducing a Narrow Space Map Recognition Strategy (NSMR), Sampling Guidance Strategy (SG), and Gravitational Elimination Strategy (GE) to address the efficiency problems of the traditional IRRT* algorithm when traversing narrow passages. NSMR identifies narrow passage regions by traversing the grid map and generating a special bias point in these regions. The SG was introduced at special bias points. When random sampling points fall within the range centered on the bias point, they quickly iterate to the specific bias point under the combined effect of gravity and repulsion, generating a large number of effective sample points, thus enabling the robot to efficiently navigate through narrow passages. The GE was used to eliminate the bias point when the robot passed. In the simulation section, data such as the optimal path length, average path length, and algorithm efficiency are analyzed. When comparing IRRT*-APF with IRRT*, the former shows significant advantages in terms of navigation efficiency and the number of sample points when traversing narrow passages.