Transparency-optimized position–position control for human-scale bilateral teleoperation

The position–position (P–P) architecture is a simple and inherently stable control strategy for bilateral teleoperation systems, but its transparency is fundamentally limited by robot dynamics, particularly in human-scale teleoperation, where large inertia and damping degrade operator perception in free motion. In this paper, we propose a transparency-optimized P–P control architecture that leverages estimated inverse dynamics to compensate for the leader and follower robot dynamics without relying on force/torque sensors. The proposed controller preserves the simplicity and stability of the P–P scheme while significantly improving transparency. The approach is implemented and experimentally validated on a human-scale WAM bilateral teleoperation system. Experimental results show that the proposed architecture substantially improves free-motion transparency, reducing joint position tracking error by up to 32% and leader-side impedance by up to 56% compared to a gravity-compensated P–P controller. In hard-contact scenarios, the proposed method maintains comparable maximum transmittable impedance while achieving up to 53% improvement in force tracking accuracy. These results demonstrate that dynamic compensation can significantly enhance the practical transparency of P–P teleoperation systems for human-scale robots without increasing hardware complexity.