Task Planning of Robot Arm Based on 3D Voxel-semantic Map of Virtual Space
Task planning algorithm is the basis of robot arm's grasping ability in unknown environment.A virtual space task planning method based on 3D voxel-semantic map is proposed.First of all,the complete point cloud scene is obtained and spliced.The Mask RCNN network is used for the object detection and instance segmentation.A voxel-semantic hybrid map composed of 3D point cloud,semantic information and 3D CAD model is constructed.Secondly,the A*algorithm is optimized to complete the optimal path planning of the robot arm effector,and the trajectory optimization is completed through the Bessel curve.Thirdly,the pose of robot arm corresponding to grasping different geometers is explored,and the complex task planning of grasping and placing is decomposed to form a semantic-driven spatial task planning.The effectiveness and rapidity of the proposed algorithm are verified by simulation experiments in virtual space.Meanwhile,the proposed algorithm can support the robot arm to perform various complex grasping tasks and improve the intelligence level of the robot arm.
万方数据
维普
