Where2comm: Communication-Efficient Collaborative Perception via Spatial Confidence Maps

TL;DR

Where2comm introduces a novel spatial confidence map that enables communication-efficient collaborative perception by focusing on critical spatial areas, significantly reducing bandwidth while improving perception accuracy across multiple scenarios.

Contribution

The paper proposes a new spatial confidence map and a framework that dynamically adjusts communication based on perceptual importance, enhancing efficiency and performance in multi-agent perception tasks.

Findings

Achieves over 100,000x reduction in communication volume.

Outperforms previous methods like DiscoNet and V2X-ViT in 3D object detection.

Effective across various modalities, agents, and datasets.

Abstract

Multi-agent collaborative perception could significantly upgrade the perception performance by enabling agents to share complementary information with each other through communication. It inevitably results in a fundamental trade-off between perception performance and communication bandwidth. To tackle this bottleneck issue, we propose a spatial confidence map, which reflects the spatial heterogeneity of perceptual information. It empowers agents to only share spatially sparse, yet perceptually critical information, contributing to where to communicate. Based on this novel spatial confidence map, we propose Where2comm, a communication-efficient collaborative perception framework. Where2comm has two distinct advantages: i) it considers pragmatic compression and uses less communication to achieve higher perception performance by focusing on perceptually critical areas; and ii) it can handle varying communication bandwidth by dynamically adjusting spatial areas involved in communication. To evaluate Where2comm, we consider 3D object detection in both real-world and simulation scenarios with two modalities (camera/LiDAR) and two agent types (cars/drones) on four datasets: OPV2V, V2X-Sim, DAIR-V2X, and our original CoPerception-UAVs. Where2comm consistently outperforms previous methods; for example, it achieves more than $100,000 \times$ lower communication volume and still outperforms DiscoNet and V2X-ViT on OPV2V. Our code is available at https://github.com/MediaBrain-SJTU/where2comm.