A scalar per patch from pre-trained ViTs enables fast moving navigation in the real world
For robotics researchers, this work provides empirical guidance on designing visual encoders for real-world navigation, though the findings are incremental as they build on known distillation and bottleneck techniques.
This large-scale study (966 episodes, 24km) evaluates pre-trained visual encoders for real-world point goal navigation, finding that multi-teacher distillation and bottlenecking encoders to a scalar per patch improve performance and yield interpretable affordance features, while finetuning on privileged information outperforms RGB-only training.
Trained policies for real-world robotics rely on computer vision components, typically in the form of pre-trained visual encoders. These encoders are an essential component and it has been shown that their power does not emerge from training on robotics downstream losses alone. Pre-training with auxiliary losses in the form of computer-vision pre-text tasks is a defining factor and heavily conditions agent performance in robotics tasks. In this unprecedented large-scale study, we ran 966 navigation episodes of static point goal navigation in a real-world building for 24km and asked which components really matter for the computer vision aspects of robotics: we evaluate state-of-the art visual encoders in realistic conditions. We explore the usefulness of heterogeneous multi-teacher distillation leading to encoders with multiple different and complementary skills. We investigate how much information from these encoders is necessary for robotics by bottlenecking them in a principled and spatially useful way and we show that this leads to the emergence of interpretable features linked to affordances. We also argue that training policies on RGB data alone does not lead to an optimal usage of visual features and show this by finetuning policies pre-trained on privileged information. All in all, we paint a more complete picture of what aspects of computer vision are relevant for real-world navigation.