G-RRT*: Goal-oriented sampling-based RRT* path planning Algorithm for mobile robot navigation with improved convergence rate

Randomized sampling-based path planning algorithms are widely used for mobile robot navigation in complex configuration space. The optimal Rapidly Exploring Random Tree (RRT*) is one of the popular sampling-based path planning algorithms that guarantees collision-free optimal path planning solutions. Even though the RRT* path planning algorithm is asymptotically optimal, its convergence is very slow. To address this problem, this paper proposes a Goal-oriented RRT* algorithm called G-RRT*. The key idea of G-RRT* is to reduce the sampling space by generating more samples near the goal configuration. The proposed algorithm is validated in a maze environment using existing algorithms. The proposed G-RRT* path planning algorithm outperforms RRT* and Informed RRT* in three performance measures convergence time, the initial cost solution, and the number of nodes visited.