Bimanual Haptics for Humanoid Robot Teleoperation Using ROS and V-REP

The high degree of complexity associated with humanoid robotic platforms significantly hinders the definition of accurate models. Robot learning based on human teleoperation provides suitable ways and means of meeting this challenge. Teleoperating the robot in different motion and balance tasks through a haptic interface, endows the user with the ability to "feel" the dynamics of the system and react to it, taking actions to maintain the robot's balance. The data collected during the demonstrations can then be used on computational learning algorithms to teach the robot how to move, balance, and walk on its own. However, the complexity of real platforms makes the control and data gathering during the teleoperation an intricate task. In order to enhance the user's control over the teleoperation, a dual haptic joystick configuration is introduced, which is first tested in a simulated model of the real platform in V-REP. This paper presents a setup solution to implement the bimanual teleoperation of a humanoid robot, based on a distributed ROS network that bridges the haptic devices and the simulator. Data gathering during several balancing tasks has shown to be possible, and the simulation's behavior is reliable enough to support the development of an infrastructure to operate the real robot based on this approach.