Bandwidth-Adaptive Feature Sharing for Cooperative LIDAR Object Detection

Situational awareness as a necessity in the connected and autonomous vehicles\n(CAV) domain is the subject of a significant number of researches in recent\nyears. The driver's safety is directly dependent on the robustness,\nreliability, and scalability of such systems. Cooperative mechanisms have\nprovided a solution to improve situational awareness by utilizing high speed\nwireless vehicular networks. These mechanisms mitigate problems such as\nocclusion and sensor range limitation. However, the network capacity is a\nfactor determining the maximum amount of information being shared among\ncooperative entities. The notion of feature sharing, proposed in our previous\nwork, aims to address these challenges by maintaining a balance between\ncomputation and communication load. In this work, we propose a mechanism to add\nflexibility in adapting to communication channel capacity and a novel\ndecentralized shared data alignment method to further improve cooperative\nobject detection performance. The performance of the proposed framework is\nverified through experiments on Volony dataset. The results confirm that our\nproposed framework outperforms our previous cooperative object detection method\n(FS-COD) in terms of average precision.\n