Point Cloud Models Improve Visual Robustness in Robotic Learners
This addresses robustness issues for robotic learners facing visual variations, though it is incremental as it builds on existing methods with a new encoding approach.
The paper tackles the problem of visual control policies degrading under changes in visual conditions like lighting or camera position, and finds that point cloud-based policies are significantly more robust than RGB-D counterparts, with the proposed Point Cloud World Model (PCWM) also improving sample efficiency during training.
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