Learning Navigational Visual Representations with Semantic Map Supervision
This addresses the challenge of enabling household robots to better perceive and navigate environments by incorporating spatial relationships, which is an incremental improvement over existing visual pre-training methods.
The paper tackles the problem of learning visual representations for robot navigation by introducing a method that contrasts egocentric views with semantic maps, transferring spatial and semantic information to improve navigation performance. The result is state-of-the-art performance on object-goal and vision-and-language navigation tasks, achieving 47% SR and 41% SPL on a test server.
Being able to perceive the semantics and the spatial structure of the environment is essential for visual navigation of a household robot. However, most existing works only employ visual backbones pre-trained either with independent images for classification or with self-supervised learning methods to adapt to the indoor navigation domain, neglecting the spatial relationships that are essential to the learning of navigation. Inspired by the behavior that humans naturally build semantically and spatially meaningful cognitive maps in their brains during navigation, in this paper, we propose a novel navigational-specific visual representation learning method by contrasting the agent's egocentric views and semantic maps (Ego$^2$-Map). We apply the visual transformer as the backbone encoder and train the model with data collected from the large-scale Habitat-Matterport3D environments. Ego$^2$-Map learning transfers the compact and rich information from a map, such as objects, structure and transition, to the agent's egocentric representations for navigation. Experiments show that agents using our learned representations on object-goal navigation outperform recent visual pre-training methods. Moreover, our representations significantly improve vision-and-language navigation in continuous environments for both high-level and low-level action spaces, achieving new state-of-the-art results of 47% SR and 41% SPL on the test server.