Point Cloud Models Improve Visual Robustness in Robotic Learners

TL;DR

This paper demonstrates that point cloud-based visual control policies enhance robustness to visual changes and improve training efficiency in robotic learning compared to RGB-D methods.

Contribution

Introduction of a novel Point Cloud World Model (PCWM) and point cloud control policies that outperform prior methods in robustness and sample efficiency.

Findings

Point cloud policies are more robust to visual variations.

PCWM significantly outperforms prior models in training efficiency.

Point cloud reasoning improves robotic visual control performance.

Abstract

Visual control policies can encounter significant performance degradation when visual conditions like lighting or camera position differ from those seen during training -- often exhibiting sharp declines in capability even for minor differences. In this work, we examine robustness to a suite of these types of visual changes for RGB-D and point cloud based visual control policies. To perform these experiments on both model-free and model-based reinforcement learners, we introduce a novel Point Cloud World Model (PCWM) and point cloud based control policies. Our experiments show that policies that explicitly encode point clouds are significantly more robust than their RGB-D counterparts. Further, we find our proposed PCWM significantly outperforms prior works in terms of sample efficiency during training. Taken together, these results suggest reasoning about the 3D scene through point clouds can improve performance, reduce learning time, and increase robustness for robotic learners. Project Webpage: https://pvskand.github.io/projects/PCWM