Obstacle Avoidance Planning And Experimental Study For Multi-redundant Manipulator Based On Computer Vision

Emerging in the late 1980s and fast developing in the 1990s, the advancedredundant robot system has become one of the most active research fields with a widerange of application in extreme environment, such as manufacture, aerospace anddeep sea explore. In recent years, redundant robot has gradually evolved to becomethe intelligent robot with a greater autonomy getting from the application of advancedsensors, especially from the maturation and perfection of computer vision technology.In this thesis, the systematic research of both a rapid image processing algorithmand an obstacle avoidance algorithm for the redundant manipulator is conducted. Theintegration of algorithms is achieved through the software programming by usingVisual C++ 6.0. A robot vision experiment platform based on servo-system is set up,on which the robot can realize the avoidance of random moving obstacles whilefollowing the end-effector trajectory at the same time.First, based on the consideration of effectiveness and real time performance, theimage processing procedure is consisting of image medium filter, image binary-conversion, Sobel operator edge detection and Hough transformation circle detectionare well designed, through which the real time positional information of the randomlymoving obstacle car can be obtained.Second, the obstacle avoidance algorithms on both position level and velocitylevel are separately conducted, and their comparison is also systematically analyzed.A new"Shortest Distance"index and a new"Obstacle Avoidance Area"index areproposed based on the gradient projection algorithm to achieve the hybrid indexobstacle avoidance planning. Simulations are implemented to prove the effectivenessof the algorithm in improving the obstacle avoidance performance of the manipulatorover its whole motion.Finally, based on the former theoretical research, an experiment system of aplanar 3R2P redundant robot vision based, including software and hardware, isaccomplished. The experiments on the system have been done to prove the feasibilityand effectiveness of the system.