This assignment mainly completed the RRT and RRT* algorithms in matlab and ROS environments, and conducted corresponding tests.
Table of contents
1. Matlab implements RRT
The sampling-based method is different from the search-based algorithm. The search-based planning algorithm is mainly suitable for path planning in low-dimensional space, which is the environment where hw arranged in the first two chapters is located, but it often occurs in high-dimensional space. Curse of dimensionality. RRT (Rapidly-exploring Random Tree rapid expansion random number algorithm) is probabilistically complete but not optimal. RRT needs to sample the entire state space and is difficult to pass through narrow channels. The efficiency is not very high. Its algorithm flow is as follows:
Simulate in MATLAB:
2. ROS implements RRT*
The RRT* algorithm is implemented in ROS based on the ompl library. RRT* is an improved version of the RRT algorithm, which mainly adds the pruning process. Through pruning, RRT* is probability complete and optimal, but RRT* still needs Sampling is performed in the entire state space. For the experimental process, please refer to: Chapter 1. The experimental results are as follows:
