Blending Primitive Policies in Shared Control for Assisted Teleoperation

TL;DR

This paper introduces a method using Dynamical Movement Primitives to blend human and robot policies in shared control, reducing human intervention in teleoperation tasks while maintaining performance.

Contribution

It proposes a primitive blending approach that implicitly combines human and robot policies without explicit weighting, enabling efficient shared control in teleoperation.

Findings

Achieved comparable task performance to conventional methods.

Reduced human intervention by over 60%.

Validated through user studies on goal reaching and obstacle avoidance.

Abstract

Movement primitives have the property to accommodate changes in the robot state while maintaining attraction to the original policy. As such, we investigate the use of primitives as a blending mechanism by considering that state deviations from the original policy are caused by user inputs. As the primitive recovers from the user input, it implicitly blends human and robot policies without requiring their weightings -- referred to as arbitration. In this paper, we adopt Dynamical Movement Primitives (DMPs), which allow us to avoid the need for multiple demonstrations, and are fast enough to enable numerous instantiations, one for each hypothesis of the human intent. User studies are presented on assisted teleoperation tasks of reaching multiple goals and dynamic obstacle avoidance. Comparable performance to conventional teleoperation was achieved while significantly decreasing human intervention, often by more than 60%.