Research On Target Recognition,Positioning And Path Planning Of Intelligent Tennis Picking Robot

With the vigorous implementation of tennis events in China,there are more and more professional tennis players and tennis enthusiasts in China.Tennis players need a lot of training every day,and repeatedly picking up tennis balls scattered on the ground greatly increases the burden on the athletes.Image recognition and path planning technologies are developing rapidly,but few of these intelligent technologies are applied to tennis ball picking up at present.People pick up tennis balls still rely largely on mechanical manual-assisted ball-picking devices.In order to solve the problem that picking up a tennis ball in the course of training is time-consuming and laborious,make Intelligent ball picking come true,this paper studies the target recognition,positioning and path planning technology of the intelligent ball picking robot.Firstly,this paper adopts the tennis recognition algorithm based on HSV color space to recognize tennis,at the same time,use Mask-RCNN to train my own data set to recogize tennis and obstacles.The experimental results of the two algorithms are compared.Finally,because the recognition technology based on color space is greatly affected by the weather,light and other factors,the method by Mask-RCNN is adopted.Through binocular camera calibration and 3D reconstruction,the coordinates of tennis balls and obstacles are obtained and output in real timeSecondly,because there are tennis and obstacles in the ground at the same time.The excellent global planning ability of the ant colony algorithm and the good obstacle avoidance performance of the artificial potential field algorithm are more in line with the actual work needs of the tennis picking robot,that is,avoiding obstacles while picking the tennis ball.Thus this paper combines the traditional ant colony algorithm and artificial potential field method to the potential field of ant colony algorithm,and its population size and pheromone distribution are improved.Multi-objective path planning based on improved potential ant colony algorithm is realized.And the simulation comparison experiment by the Matlab simulation software to show that this algorithm performs better in the selection of optimal path and the convergence rate.That verifies the feasibility of the algorithm.Finally,this paper designs the control system,including the hardware selection and the upper computer human-computer interaction program based on c# and the lower computer driver program based on STM32 MCU driver that controls the robot to pick up tennis balls.