Beyond Domain Randomization: Event-Inspired Perception for Visually Robust Adversarial Imitation from Videos

TL;DR

This paper introduces an event-inspired perception method that converts RGB videos into sparse, transient-based representations, enabling visually robust imitation learning that is invariant to appearance changes and domain shifts.

Contribution

The authors propose a biologically inspired sensory representation that focuses on temporal transients, improving robustness in imitation learning without extensive data augmentation.

Findings

Achieves invariance to appearance-based distractors

Outperforms traditional visual randomization methods

Effective across various complex control tasks

Abstract

Imitation from videos often fails when expert demonstrations and learner environments exhibit domain shifts, such as discrepancies in lighting, color, or texture. While visual randomization partially addresses this problem by augmenting training data, it remains computationally intensive and inherently reactive, struggling with unseen scenarios. We propose a different approach: instead of randomizing appearances, we eliminate their influence entirely by rethinking the sensory representation itself. Inspired by biological vision systems that prioritize temporal transients (e.g., retinal ganglion cells) and by recent sensor advancements, we introduce event-inspired perception for visually robust imitation. Our method converts standard RGB videos into a sparse, event-based representation that encodes temporal intensity gradients, discarding static appearance features. This biologically grounded approach disentangles motion dynamics from visual style, enabling robust visual imitation from observations even in the presence of visual mismatches between expert and agent environments. By training policies on event streams, we achieve invariance to appearance-based distractors without requiring computationally expensive and environment-specific data augmentation techniques. Experiments across the DeepMind Control Suite and the Adroit platform for dynamic dexterous manipulation show the efficacy of our method. Our code is publicly available at Eb-LAIfO.