Study On Robot Path Planning Based On Machine Vision

With the rapid upgrading of products,the production mode of the manufacturing industry has gradually changed from mass production and large-scale manufacturing to customization and personalization,which requires the use of highly flexible robot automated production line in manufacturing.At present,when industrial robots are dealing with new production tasks,robot path planning is mainly carried out by means of teaching.However,to obtain the robot's path in this way requires special training for operators.The path planning system is not flexible enough to adapt to frequent changes in production tasks.In view of the above situation,a kind of robot path planning method based on machine vision was proposed in this paper,which obtained the robot motion path through the method based on machine vision and human interaction.The robot path optimization model was established.And the optimization model was solved by particle swarm optimization algorithm.The main research contents are as follows:The calibration of the robot path planning vision system and the acquisition of obstacle position information were studied.First,the vision system camera internal calibration and the hand-eye calibration between the camera and the robot were completed.Then,the obstacles were located based on the template matching algorithm and the SURF-based algorithm.And the position information of the obstacles in the robot base coordinate system was obtained,which provided the obstacle position information for the robot path planning.The method of robot initial path acquisition was studied based on machine vision and human-computer interaction.Track the path of the human hand through the camera and the Mark point.On this basis,in order to avoid the error caused by the change of finger joint angle during the human hand demonstration,a robot initial path acquisition method based on the human hand gripper was proposed.The experimental results showed that the initial path based on the manual gripper method was smoother.The initial path of the robot based on machine vision and human-computer interaction was optimized.The corresponding robot path optimization model was established.And the optimization model was solved by Particle Swarm Optimization(PSO).Finally,the experimental research of robot path planning was carried out.First,a robot path planning system including robot and machine vision was built.Then,the initial path acquisition and optimization of the robot were carried out in three different experimental scenarios,and the optimized robot path was experimentally verified on the robot simulator and the robot.The experimental results showed that the robot path optimized by PSO could effectively avoid obstacles and were shorter than the initial path of the robot based on the human hand demonstration,thus improving the efficiency of industrial production.