Human Robot Dual-Manipulators Active Motion Transfer Method Reshearch Based On Virtual Reality

Recently, with the increasing attention of the dual-arm humanoid robot by people, it has been hoped to replace human to complete varied kinds of complex tasks for its dexterity, and won’t increase the burden of programming. At the same time, with the development of the intuitive human robot interaction technology, which provides a new method for the motion control problem of dual-arm humanoid robot. This research aims at the study of human robot dual-manipulators active motion transfer method research based on virtual reality, where operator can operate the robot to complete complicated tasks in unstructured and uncertain environment intuitively and efficiently and needs no requirements of professional skills and trainings. This research mainly completed the following works.In the process of controlling robot to perform operation, human must know the operation task and its real-time updated state clearly. So human arm motion transfer is not blind but based on the visual feedback subsystem constructed by using virtual reality technology. Firstly, the paper used the Oculus Rift sensor to provide the first person perspective of Baxter to the operator in real-time that is transferring the scene observed by robot to the operator in real-time based on virtual reality technology, which will let the operator completely immersed in the robot’s environment to achieve more intuitive and efficient operation.Secondly, the manipulator’s kinematics model of the human arm is established. Considering the kinematic model of Baxter dual-manipulators and the flexible requirements of human arm, using the 3D position of shoulder, elbow, wrist and hand point captured by Kinect sensor, this paper took a 7 DOFs series manipulator model as the D-H model of the human upper limb, and proposed a geometric vector method to calculate the joint angle of human upper limb to achieve the purpose of human arm motion transfer.Thirdly, this paper proposed the human and robot arm movement transfer method based on the Cartesian position and orientation mapping method. Considering the difference configuration of human arm and Baxter manipulator, we use linear space mapping algorithm for the space mapping of human upper limb and robot manipulator to get the Baxter manipulator end-effector pose. Then, using the inverse kinematics solution to calculate the joint angles for motion transfer. In order to solve the problem of multiple solutions for redundant manipulator, the paper proposed a numerical inverse solution generating humanoid motion based on CLIK algorithm.In the end, the human dual-manipulators active motion transfer system verification platform based on virtual reality is built, through which realizing the control of Baxter dual-manipulators, completing a series of operation tasks, and verifying the effectiveness and stability of our methods.