Spatially Visual Perception for End-to-End Robotic Learning

TL;DR

This paper presents a video-based spatial perception framework using 3D representations and novel augmentation to improve robustness in robotic learning across diverse camera conditions.

Contribution

It introduces a new spatial perception system combining 3D representations, AugBlender augmentation, and monocular depth estimation for better generalization.

Findings

Significantly improved success rate across diverse camera exposures

Enhanced robustness against lighting changes in dynamic environments

Demonstrated potential for scalable, low-cost embodied intelligence solutions

Abstract

Recent advances in imitation learning have shown significant promise for robotic control and embodied intelligence. However, achieving robust generalization across diverse mounted camera observations remains a critical challenge. In this paper, we introduce a video-based spatial perception framework that leverages 3D spatial representations to address environmental variability, with a focus on handling lighting changes. Our approach integrates a novel image augmentation technique, AugBlender, with a state-of-the-art monocular depth estimation model trained on internet-scale data. Together, these components form a cohesive system designed to enhance robustness and adaptability in dynamic scenarios. Our results demonstrate that our approach significantly boosts the success rate across diverse camera exposures, where previous models experience performance collapse. Our findings highlight the potential of video-based spatial perception models in advancing robustness for end-to-end robotic learning, paving the way for scalable, low-cost solutions in embodied intelligence.