Deep Reinforcement Learning for Single-Shot Diagnosis and Adaptation in Damaged Robots

TL;DR

This paper introduces a deep reinforcement learning approach that enables damaged robots to diagnose their damage and adapt their gait in a single trial, using a damage-aware control architecture with minimal computational complexity.

Contribution

It presents a novel damage-aware control system that diagnoses damage and adapts policies with a single policy trained across diverse damages, enhancing robustness in robotic control.

Findings

Single policy adapts to various damages effectively

LSTM-based damage diagnosis predicts damage type accurately

Robust policy learned through domain randomization

Abstract

Robotics has proved to be an indispensable tool in many industrial as well as social applications, such as warehouse automation, manufacturing, disaster robotics, etc. In most of these scenarios, damage to the agent while accomplishing mission-critical tasks can result in failure. To enable robotic adaptation in such situations, the agent needs to adopt policies which are robust to a diverse set of damages and must do so with minimum computational complexity. We thus propose a damage aware control architecture which diagnoses the damage prior to gait selection while also incorporating domain randomization in the damage space for learning a robust policy. To implement damage awareness, we have used a Long Short Term Memory based supervised learning network which diagnoses the damage and predicts the type of damage. The main novelty of this approach is that only a single policy is trained to adapt against a wide variety of damages and the diagnosis is done in a single trial at the time of damage.