Vision-based Robust Robot Path Planning Algorithms

Path planning is one of the research focuses in the field of mobile robots.With the rapid development of industrial robots,human-machine collaboration has gradually become the main working mode of modern industrial robots.In the human-machine collaboration system,the interaction between human workers and robots is mainly security-oriented robot dynamic path planning.When the robot moves,it is required to be subjectively active in a dynamic and uncertain environment,as well as to be safe and efficient.In view of the above analysis,path planning in the human-machine collaboration system brings the following challenges: accurately identifying the safe area in the environment and converting it into an environmental model;planning the path for safety and efficiency in a known environment;and correcting the path effectively when a potential collision is detected.In order to solve the above problems,this paper mainly studies a vision-based flexible path planning algorithm for robots.The specific contents are as follows:First,the geometric model was selected as the environmental model for this study.Then,a vision-based method is proposed to perform safe area recognition on environmental video frame images.A convolutional neural network is used to establish a semantic segmentation model of the encoding and decoding structure.Then,due to the uniqueness of the indoor environment,image data is collected to establish a training set and the model is trained.Finally,the edge detection technology is used to process the video frame image,and the result is combined with the semantic segmentation result to make the result more accurate.Finally,the algorithm of rapidly-exploring random trees based on randomness and RRT* algorithm with progressive optimality are introduced,and the improvement is made for its randomness and the path containing more redundant nodes.Then it analyzes the requirements for safety and efficiency in a dynamic environment,and proposes an improved RRT* and artificial potential field hybrid dynamic path planning algorithm.The algorithm first uses the improved RRT* algorithm to obtain the initial path,when a potential collision is detected,the path is quickly modified according to the principle of the repulsion field of the artificial potential field to obtain the next path,and the improved RRT* is used to obtain the remaining path.The design experiment is carried out using python language in Anaconda environment.Through simulation experiments,it is verified that the improved RRT * and artificial potential field hybrid dynamic path planning algorithm can effectively deal with various obstacles and enable the robot to actively avoid obstacles when it detects potential collisions.Therefore,the improved RRT* and artificial potential field hybrid dynamic path planning algorithm meets the requirements of human-machine cooperative factories for mobile robots.