Path planning using Goal Biased Gaussian Distribution based RRT*

Rapidly Exploring Random Tree (RRT) is a sampling based heuristic path planning approach used. An extended version of RRT, i.e., optimal RRT (RRT*), is widely used for path planning due to its asymptotic convergence and single query properties to find the optimal path for autonomous vehicles and robots. As contemporary autonomous vehicles demand more accurate and fast path planning algorithms, further improvements of RRT* are needed for safe and superior operations. This research introduces a new approach in the RRT* algorithm to find an optimal path in less computational time without compromising the useful characteristics of the RRT*. The proposed algorithm uses the Goal Biasing methodology that focuses on tree expansion towards the end point and Gaussian Distribution, which helps to reduce the overall state space. The proposed algorithm is a modification in RRT* for implementation of the goal biasing and Gaussian distribution methodologies. The proposed algorithm was evaluated on five different illustrative environments with different obstacle characteristics for its efficiency. The numerical experiments demonstrate that the proposed algorithm is more efficient regarding the distance cost for the path and computational time, when compared to RRT*