Natural Gradient Shared Control

TL;DR

This paper introduces a formalism for shared control in robotics using natural gradients, enabling a balance between user authority and autonomous support, demonstrated through improved task efficiency in a user study.

Contribution

It presents a novel shared control framework based on natural gradients and Fisher information approximation, enhancing user autonomy while supporting task performance.

Findings

Improved task completion efficiency in user study.

Maintains user control authority effectively.

Outperforms baseline shared control methods.

Abstract

We propose a formalism for shared control, which is the problem of defining a policy that blends user control and autonomous control. The challenge posed by the shared autonomy system is to maintain user control authority while allowing the robot to support the user. This can be done by enforcing constraints or acting optimally when the intent is clear. Our proposed solution relies on natural gradients emerging from the divergence constraint between the robot and the shared policy. We approximate the Fisher information by sampling a learned robot policy and computing the local gradient to augment the user control when necessary. A user study performed on a manipulation task demonstrates that our approach allows for more efficient task completion while keeping control authority against a number of baseline methods.