A Data-Driven Motion Mapping Method for Space Teleoperation of Kinematically Dissimilar Master/Slave Robots

Abstract

There is demand to develop methods for the master/slave teleoperation robots with dissimilar structures. However, the challenge in this case is to establish a suitable motion mapping, as their kinematics are non-homothetic. To solve this problem, this paper presents a data-driven method that enables the implementation of teleoperation system through kinematic transformation at joint level. This approach leverages the power of Drift Control, Gaussian Mixture Model (GMM) and proposed Coupling Gaussian Mixture Regression (CGMR) to encapsulate the statistical relation of the motion mapping from the small workspace of a haptic device to large workspace of an industrial robot. Here, Drift Control is adopted to progressively relocate the physically limited workspace of the master/slave robots. And the statistical relation of the motion mapping is established using GMM and proposed CGMR in order to solve the problem of multiple solutions for forward kinematics (FK) of the master robot and inverse kinematics (IK) of the slave robot. The experimental results show the feasibility of the teleoperation system using a 6-DOF parallel manipulator (6-P-2P-S) as a master robot and a 6-DOF industrial robot (OGD-JLRB20) as a slave robot. Also, the proposed controller shows good tracking to span the slave workspace.