Adaptive Motion Generation Using Uncertainty-Driven Foresight Prediction

TL;DR

This paper presents an adaptive robot control method that uses uncertainty-driven foresight prediction to generate flexible motions in uncertain environments, demonstrated through a door opening task with multiple possible actions.

Contribution

It extends predictive learning-based control with a foresight module that samples futures to minimize uncertainty, enabling adaptive behavior in unstructured scenarios.

Findings

The model successfully adapted to different door opening methods.

It outperformed conventional methods in stability during adaptation.

Lyapunov analysis showed the model embeds uncertainty in its policy.

Abstract

Uncertainty of environments has long been a difficult characteristic to handle, when performing real-world robot tasks. This is because the uncertainty produces unexpected observations that cannot be covered by manual scripting. Learning based robot controlling methods are a promising approach for generating flexible motions against unknown situations, but still tend to suffer under uncertainty due to its deterministic nature. In order to adaptively perform the target task under such conditions, the robot control model must be able to accurately understand the possible uncertainty, and to exploratively derive the optimal action that minimizes such uncertainty. This paper extended an existing predictive learning based robot control method, which employ foresight prediction using dynamic internal simulation. The foresight module refines the model's hidden states by sampling multiple possible futures and replace with the one that led to the lower future uncertainty. The adaptiveness of the model was evaluated on a door opening task. The door can be opened either by pushing, pulling, or sliding, but robot cannot visually distinguish which way, and is required to adapt on the fly. The results showed that the proposed model adaptively diverged its motion through interaction with the door, whereas conventional methods failed to stably diverge. The models were analyzed on Lyapunov exponents of RNN hidden states which reflect the possible divergence at each time step during task execution. The result indicated that the foresight module biased the model to consider future consequences, which lead to embedding uncertainties at the policy of the robot controller, rather than the resultant observation. This is beneficial for implementing adaptive behaviors, which indices derivation of diverse motion during exploration.