An Improved RRT Robot Autonomous Exploration and SLAM Construction Method

When the indoor environment is unknown, how to make robots carry out effective autonomous exploration and construct related maps is one of the key issues in the field of mobile robots. In view of the fact that the real-world environment is usually partially observable and uncertain, an improved method of autonomous exploration and SLAM construction based on Rapid-exploration Random Tree (RRT) is proposed. In the local RRT exploration part, the autonomous exploration problem is regarded as a partially observable Markov decision process (POMDP) problem. Boundary points are extracted in the boundary area of known space and unknown space, and the robot is directed to the unexplored area. At the same time, during the exploration process, the global RRT tree with adaptive step values is established to explore the boundary points at the far end of the robot. The two methods are merged to speed up the search for boundary points. After the best advantage is obtained, the robot is continuously moved to the best advantage through closed-loop control. Based on the original construction of a 2D grid map, a “parallel construction” idea was proposed to construct a 3D octree map at the same time. The effectiveness of the proposed method is verified by simulation experiments and actual scenarios in a robot operating system (ROS).