Surgical Robot Transformer (SRT): Imitation Learning for Surgical Tasks

TL;DR

This paper presents a novel imitation learning approach for surgical robots that leverages approximate kinematics data, enabling effective training and execution of fundamental surgical tasks without extensive data correction.

Contribution

Introduction of a relative action formulation that allows imitation learning on the da Vinci robot using approximate kinematics data, facilitating direct use of clinical data for surgical robot training.

Findings

Successful execution of tissue manipulation, needle handling, and knot-tying tasks.

The approach enables training with existing clinical data without kinematic corrections.

Demonstrates robustness of the method in real surgical scenarios.

Abstract

We explore whether surgical manipulation tasks can be learned on the da Vinci robot via imitation learning. However, the da Vinci system presents unique challenges which hinder straight-forward implementation of imitation learning. Notably, its forward kinematics is inconsistent due to imprecise joint measurements, and naively training a policy using such approximate kinematics data often leads to task failure. To overcome this limitation, we introduce a relative action formulation which enables successful policy training and deployment using its approximate kinematics data. A promising outcome of this approach is that the large repository of clinical data, which contains approximate kinematics, may be directly utilized for robot learning without further corrections. We demonstrate our findings through successful execution of three fundamental surgical tasks, including tissue manipulation, needle handling, and knot-tying.