Humanoid Loco-Manipulation of Pushed Carts Utilizing Virtual Reality Teleoperation

Abstract Material handling continues to be a large body of research for humanoid robots and loco-manipulation. Cart pushing remains a topic of interest in research and industry, yet attempts to automate a large range of different carts has proved difficult without advanced grasping techniques. To fill this gap, a method utilizing a VR interface for grasping of the cart is proposed. In contrast to previous techniques using a capture point control method, whole body inverse kinematics are used to compensate for additional cart dynamics, along with arm compliance for grasping the cart. Experiments include an operator attempting to grasp a cart that the robot has never encountered, and then walking and steering this cart using the DRC-Hubo humanoid robot. Results are presented analyzing the ZMP of the robot, as well as whole-body kinematics simulation and VR interface validation.