Redundancy-aware Action Spaces for Robot Learning

TL;DR

This paper introduces ER, a novel action space formulation for robot arms that combines the control precision of joint space with the data efficiency of task space, improving learning efficiency and control in manipulation tasks.

Contribution

The paper proposes ER, a new redundancy-aware action space that leverages manipulator redundancies to unify the advantages of joint and task space control, with two implementations and validated results.

Findings

ERJ outperforms existing methods in multiple settings.

ER improves learning efficiency in simulated and real environments.

ER provides fine-grained control over robot configurations.

Abstract

Joint space and task space control are the two dominant action modes for controlling robot arms within the robot learning literature. Actions in joint space provide precise control over the robot's pose, but tend to suffer from inefficient training; actions in task space boast data-efficient training but sacrifice the ability to perform tasks in confined spaces due to limited control over the full joint configuration. This work analyses the criteria for designing action spaces for robot manipulation and introduces ER (End-effector Redundancy), a novel action space formulation that, by addressing the redundancies present in the manipulator, aims to combine the advantages of both joint and task spaces, offering fine-grained comprehensive control with overactuated robot arms whilst achieving highly efficient robot learning. We present two implementations of ER, ERAngle (ERA) and ERJoint (ERJ), and we show that ERJ in particular demonstrates superior performance across multiple settings, especially when precise control over the robot configuration is required. We validate our results both in simulated and real robotic environments.