Abstract
Correspondence identification (CoID) is an essential capability for multi-robot collaborative perception, which allows a group of robots to consistently refer to the same objects in their own fields of view. In real-world applications, such as connected autonomous driving, connected vehicles cannot directly share their raw observations due to the limited communication bandwidth. To address this challenge, we propose a novel approach of bandwidth-adaptive spatiotemporal CoID for collaborative perception, where robots interactively select partial spatiotemporal observations to share with others, while adapting to the communication constraint that dynamically changes over time. We evaluate our approach over various scenarios in connected autonomous driving simulations. Experimental results have demonstrated that our approach enables CoID and adapts to the dynamic change of bandwidth constraints. In addition, our approach achieves $8\%$-$56\%$ overall improvements in terms of covisible object retrieval for CoID and data sharing efficiency, which outperforms the previous techniques and achieves the state-of-the-art performance.
