Research On Path Planning Method Based On Case-based Reasoning For Mobile Robot

Mobile robot is an important part of research in the field of robotics. How to find a safe andcollision-free path in unstructured environments is always an important issue in path planning ofmobile robots. By the summaries and analysis of the past path planning methods, we could findthat global path planning in static environment is relatively mature. But in real environment, theresearch on the autonomous navigation in unstructured dynamic environment has more practicalvalue, so further research is of great significance. Because the environment around the robot isunknown and uncertain, autonomous navigation becomes hardly to achieve.One of the main issues for mobile robot to navigate in uncertain environments is the lack ofprior information about them. If the robot could simulate the human’s thinking, make use of thepast experience to find the similar case, and reuse the solution to solve the current problem, itcould greatly improve the performance of mobile robots in time and distance.This paper has proposed a new path planning method which combined case-based reasoningtechnique with the modified potential field method. We use case-based reasoning to get theavailable prior information by retrieving the past cases and revising the solutions to solve thecurrent problems. Case-based reasoning is an incremental and sustainable approach in learning,continually update case base could solve the problem in different scenarios. The experimentalresults demonstrate this method not only avoids obstacles effectively and safely, but also greatlyimproves the performance of the robot in terms of time and distance of the path taken from thestart to the target.To validate the feasibility of this algorithm, the open-source robot operating system ROS isused as the software platform and the omnidirectional mobile robot is used as the experimentalrobot. Moreover, the performance of the improved algorithm is verified、tested、analyzed andevaluated in this paper. Firstly, we take kinect as the vision sensor to extract environmentalcharacteristics in real time and obtain the information of obstacles around the robot, then transferthe information to the computer. Secondly, map the point cloud into a2-dimensional continuousspace. Finally, cluster the points of2D-map by DBSCAN algorithm and detect the edges of theobstacles based on canny operator to obtain the useful information which is needed in path planning. The results of the experiments show that this method can generate an optimal path andsuccessfully achieve autonomous avoidance for mobile robot.