Inertia Compensation Using Flywheels in Parallel Robots for the Assisted Manipulation of Large Payloads

This paper proposes the use of reaction wheels in parallel mechanisms for physical human-robot interaction during the co-manipulation of large payloads. The concept combines the advantages of a mechanically backdrivable robot - for hands-on-payload interaction - with the reactiveness of flywheels for the compensation of inertial loads, thereby leading to a smooth and low-inertia rendering. In the proposed approach, gravity compensation and dynamic compensation are partitioned and assigned to two subsets of actuators, namely the backdrivable joint actuators and the flywheel actuators, the latter being smaller and properly geared actuators to benefit from faster dynamics for interaction stability purposes. Simulation results of a human interaction with a planar robot to displace a payload show that the desired dynamic behaviour of the moving platform is correctly rendered, while indicating that the inertia compensation torques may vary more quickly than the gravity torques, which supports the proposed idea. Experiments are also conducted to validate the rendering of the desired virtual dynamics to the user.