Mirrorcomm cuts data use in cooperative perception by predicting updates
MirrorComm: A Communication-Efficient Cooperative Perception Framework via Mirrored World Models
Computer Vision and Pattern Recognition
Summary
Vehicles working together to sense their surroundings typically share lots of data, which can overload communication systems. The authors devised a new method called MirrorComm that lets connected vehicles predict what others are sensing, so only unexpected changes need to be sent. This prediction-based sharing reduces the amount of data sent by up to 66 times without losing accuracy in detecting objects or understanding scenes. Their tests show the system works well and that its parts each help improve efficiency.
What this means in practice
- •For automotive engineers: Implement communication-efficient data sharing in connected vehicles to reduce bandwidth when exchanging sensor information for better cooperative perception.
- •For robotic system integrators: Use mirrored world models to enable multi-robot teams to share sensor updates efficiently by sending only changes instead of full data frames.
Tested on one dataset.
Authors
Shenghan Luo, Yin Xu, Cixiao Zhang, Yang Wang, Yujie Cao, Zhuoran Xiao, Wenjun Zhang
Abstract
Cooperative perception (CP) requires substantial communication bandwidth to exchange sensing information among connected agents. A variety of compression techniques have been proposed to reduce transmission costs while retaining task-relevant information. However, most existing methods follow a frame-wise compression paradigm, overlooking temporal predictability and repeatedly transmitting information that could be inferred from historical context. To address these limitations, we propose MirrorComm, the first world model-based communication-efficient CP framework. Specifically, mirrored world models at the transmitter and receiver predict the transmitter's current observation from a shared history of its reconstructed observations. The transmitter compares this prediction with its current observation and selectively transmits observation updates to correct prediction errors, reducing repeated transmission of predictable information. Experiments on the OPV2V dataset across two downstream tasks including object detection and semantic segmentation show that MirrorComm reduces communication bandwidth requirements by up to approximately 66-fold while maintaining comparable performance to a state-of-the-art (SOTA) method. Ablation studies further demonstrate the effectiveness of the individual components.