Research On Path Planning And Target Grasping Technology Of Mobile Manipulation Robots

In recent years,with the increasing demand of the society for robots,it has become a trend to develop and research the technologies of mobile manipulation robot.Mobile manipulation robot is the combination of mobile platform and manipulator.It combines the maneuverability of manipulator and mobility of mobile platform.The security of moving paths and the ability of object recognition are the key technical indicators to measure the stability and efficiency of mobile manipulation robot in indoor environment.Therefore,this paper mainly focuses the main technologies of path planning,target recognition and motion planning of mobile manipulation robot.For the path planning technology of mobile manipulation robot,this paper firstly analyzes the rapidly-exploring random trees,and studies the related improved algorithm.On this basis,the A-RRT* algorithm based on expansion radius is proposed,which uses artificial potential field to guide the fast expansion random tree to expand to the target point.The lack of guidance of RRT* algorithm in search process is solved.The expansion radius idea is introduced to make the mobile robot avoid the narrow channel in the narrow environment and improve the safety of the mobile robot.Finally,the optimal path is obtained by smoothing the fusion algorithm.Compared with traditional RRT and RRT* algorithm,the fusion algorithm can search faster,and the path cost and smoothness can be further optimized.For the research on the target grasping technology of mobile manipulation robot,this paper firstly analyzes the camera calibration principle,establishes the conversion relationship between each coordinate system,obtains the camera’s internal parameters through Zhang’s calibration algorithm,obtains the hand-eye calibration parameters according to the installation position of the camera.The improved YOLOV4 algorithm was used to obtain the pixel coordinates of the target.By introducing the idea of Focal loss,Focal loss was used to replace the cross entropy loss and participate in the calculation of the model loss,making the model more inclined to learn positive samples and hard-to-classify samples,thus improving the detection accuracy of the target.The experimental results show that the improved algorithm effectively alleviates the error detection and missing detection problems of the original algorithm.For the motion planning problem of the manipulator,this paper adopts the seventhdegree polynomial trajectory planning algorithm to carry out the motion planning of the manipulator,and completes the 6-DOF manipulator grasping task experiment.The simulation results show that each joint Angle of the manipulator runs smoothly without violent chattering,which verifies the effectiveness and reliability of the proposed trajectory planning algorithm.Based on the above work,the function configuration of the mobile manipulation robot is completed in the ROS environment,and autonomous navigation,target recognition and grasping experiments are carried out.Experimental results show that the mobile manipulation robot designed in this paper can safely reach the target point,accurately identify the object and complete the grasping task,meet the design requirements.