Research On Vision-based Intelligent Collision Avoidance And Cooperative Operations Of Industrial Manipulators

The real-time obstacle-free path planning of a multi-DOF(degree-of-freedom)serial industrial manipulator has been a hot spot in the field of robotics research.In general,tasks performing of the traditional industrial manipulator in the complicated work environment are usually taught offline,which is incapable of perceiving the changes of the surroundings,if dynamic obstacles enter the work environment such as people or other moving obstacles,casualties or damages may be caused to the running manipulator or people who entered accidentally.To solve this problem,we proposed an intelligent collision-free path planning method for industrial manipulator which based on the robotic vision.To achieve the collision avoidance goal of path planning if dynamic moving objects appear in the work environment,the path planning of the industrial manipulator will figure out a path that avoid the obstacles.The main content of this thesis is as follows.Firstly,we analyze the kinematics of 6-DOF industrial manipulator including forward kinematics and inverse kinematics.The derivation of the solution is specified,the inverse kinematics is solved by geometric and algebraic method and closed-form solutions of the manipulator that satisfy the Pieper principle are obtained.Secondly,we proposed a dynamically moving object detection method based on the Kinect V2 sensor including the calibration of Kinect sensor and three frames difference algorithm as well as the obtaining of the point cloud of the detected moving object.Meanwhile,we proposed a intelligent collision-free path planning method for the industrial manipulator,we first analyzed the collision detection method of the industrial manipulator,compared 3 different bounding box model and eventually employ the AABB(Axis-aligned Bounding Boxes)as the bounding box model for our industrial manipulator.Then,we specified the SPARS algorithm which is aimed to construct a sparse roadmap,SPARS algorithm guarantees the asymptotically optimality of the path and the quality of the found path has theoretical boundary.The searching of shortest path and roadmap maintenance is achieved by the cost-bumped LazyPRM algorithm,which is capable of keeping the manipulator away from the collision locations since it's heuristically piloted by the cost of vertices along the candidate path.In order to verify the effectiveness of the proposed method,we used OMPL(Open Motion Planning Library)and the simulation system we developed to perform experiments,the results show that the method we proposed outperforms the traditional RRTConnect and PRM algorithms and exhibits satisfactory performance in term of both computation time cost and memory consumption upon the intelligent collision-free path planning problem of industrial manipulator.Thirdly,we present cooperative work methods of industrial manipulator cooperated with five-fingered dexterous robot hand,in the meantime,analyze the forward kinematics of five-fingered dexterous robot hand.In order to control the dexterous hand,the proposed methods use 5DT data glove,the calibration of the 5DT data glove is specified.At last,the cooperative operations are presented through the simulation system.Finally,we present the path planning system we developed including all the algorithms of the methods mentioned in this thesis and simulation system which we used to verify the correctness and effectiveness of the algorithms.