Research On Human-machine Intelligent Integration Based Path Planning For Mobile Robots

From the viewpoint of the application demands of Unmanned Combat Platform and Lunar Rover, path planning is the primly important problem for navigation and control of mobile robots in complex unstructured environments. Due to the limitations of the development of related technologies and robot system, the navigation and control of mobile robots can not be fully autonomous. In this paper, a new path planning algorithm—Artificial-guided Rapidly-Exploring Random Trees (AG-RRTs) is proposed, which integrates human being's decision to address above problem. Furthermore, a tele-control system of multi-pattern human-machine intelligent integration is designed to assist the mobile robots to realize navigation and control in complex unstructured environments.The main contributions and innovations are as following:(1) A new global path planning algorithm—AG-RRTs is proposed in this paper which is based on the intelligent integration of human and machine. In this algorithm, the guided waypoints are selected by user to identify narrow regions in the map. The planner creates multiple trees from the initial point, goal point and the waypoints, respectively. The trees iteratively extend to merge with each other to find a solution path. The proposed algorithm extends the dual-RRTs to multiple trees to promote the performance of the RRT algorithm under human interference. The effectiveness of the proposed algorithm is verified through the experiments on global path planning of indoor mobile robots.(2) Three kinds of local path planning method and corresponding tele-control patterns based on the intelligent integration of human and machine are designed and implemented in this paper. According to different combination forms of human being's intelligent and robot's autonomy, three kinds of semi-autonomous local path tele-control patterns are introduced, including: key waypoints tele-control, trajectory-guided-point tele-control, and control-guided-point tele-control. The performance and the application scopes of each method are analyzed through a series of simple obstacle avoidance experiments.(3) A tele-control system of human-machine intelligent integration is built on the mobile robot platform Pioneer3-AT. It integrates multiple tele-control patterns, including: master-slave pattern, autonomous pattern and semi-autonomous pattern. Users can interactively choose the operation pattern to meet the demands of the mobile robots with different tasks in different environments. In the synthesized experiments, the users perform the navigation task of the mobile robot in a complex, uncertain environment through different tele-control patterns, which verifies the utility of the system.