Research And Application Of Dual-machine Robot Desktop Cleaning Algorithm Based On Machine Vision

With the development of science and technology and the progress of society,the demand for service-type cleaning robots is increasing in a society of fast-paced and stressful work.Manual cleaning is often used in the desktop cleaning of restaurants,cafes,hotels,homes and enterprises,and is inefficient,costly,tedious and subjective.Therefore,the application of robots in the field of cleaning is more and more extensive.However,robots still have problems such as difficulty in detecting desktop dust,non-uniformity of cleanliness judgment,mechanical full coverage cleaning,and inability to arrange desktop items in an orderly manner In the field of desktop cleaning,resulting in low cleaning efficiency and low resource utilization.In view of the above problems,a desktop cleaning system based on machine vision technology and robot technology had been designed in this thesis.It successfully implements the functions of orderly placement and cleaning of desktop items,dust detection,cleaning area and placement area planning.The main research and work are as follows:(1)The dust detection algorithm research verification platform and a desktop cleaning experiment verification platform were established(2)The camera at the end of the left arm of the Baxter robot was calibrated,and HSV color space conversion,binarization,morphological closing operation,contour extraction were performed on the image captured by the camera.And the size,center position and rotation angle of desktop items were successfully obtained by using polygon approximation algorithm and minimum external rectangle algorithm.(3)The planning algorithm for the cleaning area and the placement area were designed,and the orderly placement and dust cleaning of the desktop items were realized successfully.(4)Enhance the texture information of dust by image equalization and Laplace transform to solve the problem that the dust information of desktop is not obvious,and extract the parameters of dust texture features by using the gray level co-occurrence matrix algorithm.Furthermore,an image similarity algorithm based on improved Canberra distance was proposed to solve the problem of illumination variation interference during dust detection.It improves the accuracy of dust detection in illumination changes,and achieves accurate detection of desktop dust.(5)The path-planning of Baxter robot motion during the capture and placement was proposed to realize the cooperation of the two arms of the robot.Thereby the collision between the desktop item and the robot arm was reduced effectively.A desktop cleaning mode with different items was designed,and the robot's crawling,placement,and movement paths in the cleaning area were planned and relevant experimental verification was performed.The results show that the cleaning effect is good,and the flexibility and intelligence of the robot in the field of desktop cleaning are effectively improved.Baxter robots for desktop cleaning was used then the algorithms for desktop dust detection,desktop cleaning,robot path planning were verified.And the function of the robot's object detection and identification,grab and place,cleaning area and placement area planning,clean area dust detection and dust cleaning was unified.The robot's cleaning of the desktop has been completed.