Development Of Intelligent Tennis Robot Based On Visual Recognition

With the development of tennis,more and more people have become tennis enthusiasts.However,the picking up of tennis balls scattered on the ground during tennis training greatly affected enthusiasts' enthusiasm.In order to solve the difficult problem of picking up tennis balls,this article developed a intelligent tennis robot based on visual recognition technology.It can intelligently identify tennis balls and pick them up into a ball basket.This article divides the robot system into three modules: a tennis recognition module,a tennis picking module,and a move control module.In tennis recognition,two tennis recognition algorithms based on tennis color characteristics,tennis outline characteristics and height information are proposed.The tennis recognition algorithm based on tennis color features mainly uses tennis HSV color features for tennis recognition.By analyzing the HSV color space of the tennis ball in the picture,the range of each channel of the tennis HSV is determined,and the picture is processed using the obtained HSV range to obtain a binary image of the tennis ball and the background.The contour image is extracted from the binary image,the minimum bounding circle for each contour is obtained,and the radius of the minimum bounding circle is screened using a predetermined tennis radius to complete the tennis recognition.The tennis recognition algorithm based on tennis color features is stable in places where light intensity is weak,but has poor robustness in outdoor environments where light intensity is high.Therefore,a tennis recognition algorithm based on tennis outline features and height information with higher robustness is proposed.The tennis recognition algorithm based on tennis outline features and height information uses two cameras to construct a stereo coordinate system.Based on the non-linear camera model,the programming program completed camera calibration,stereo-calibration,stereo rectification,stereo correspondence and coordinate system conversion,to obtain the three-dimensional coordinates of tennis in the specified robot coordinate system.The Hough circle transformation first identifies the circular contours in the image,obtains all circular contours that may be tennis balls,then uses tennis heights to filter,discarding disturbing circular contours(such as tennis shadows),and filtering to real tennis balls.The test verifies that the tennis recognition algorithm based on the outline characteristics and height information of tennis balls is stable in both indoor and outdoor operations,tennis recognition is accurate,real-time performance is good,and robustness is high.In the aspect of tennis picking mechanism,an efficient shuttlecock mechanism based on belt drive was designed.In order to reduce the complexity of the mechanism,the shuttlecock mechanism and the lifting mechanism were integrated.The overall structure of the robot uses a profile-based frame design,which makes the appearance of the robot simpler while controlling the weight of the robot.In addition,a storage ball basket,a tensioning mechanism,etc.are also designed.These mechanisms combine to form a smooth and efficient ball rolling mechanism.In the aspect of motion control system,a differential motion system with two drive wheels and two support wheels was built,and the software and hardware parts of the motion system were designed.The STM32 microcontroller is selected as the micro-controller of the motion system in the hardware part.The appropriate motor and driver are selected according to the robot load and speed requirements.The software based on the PID algorithm to control the motor speed,put forward a straight-line walking algorithm and point-to-point walking algorithm suitable for tennis court applications.After the field test on the tennis court,the robot robot designed was accurate in identification,smooth in movement,and good in picking up tennis.