# BOLT: Online Lightweight Adaptation for Preparation-Free Heterogeneous Cooperative Perception

Source: [BOLT: Online Lightweight Adaptation for Preparation-Free Heterogeneous Cooperative Perception](<https://www.sota2.com/research/paper/69f82816c0f56ed843088ac6>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-05-05T05:00:04.460Z

Authors: Kang Yang, Tianci Bu, Peng Wang, Deying Li, Yongcai Wang

Published: 2026-05-01T04:53:50.000Z

## Abstract

Most existing heterogeneous cooperative perception methods depend on prior preparation like offline joint training or tailored collaborator-model adaptation. Such preprocessing is, however, generally impractical in real scenarios, as agents are usually independently trained by different developers and meet occasionally online. This work investigates \\emph{preparation-free heterogeneous cooperative perception}, where agents use independently trained single-agent detectors without any pre-deployment coordination. We find direct cross-agent fusion under this setting greatly underperforms ego-only perception. We present BOLT, a lightweight plug-and-play module that adapts neighboring features online via ego-as-teacher distillation, requiring only ego predictions without ground-truth labels. BOLT leverages high-confidence ego perception features to guide cross-agent feature-domain alignment, while enabling neighbors to contribute features in the ego's low-confidence regions. With only 0.9M trainable parameters, BOLT improves AP@50 by up to 32.3 points over vanilla unadapted fusion in the preparation-free setting. It consistently outperforms ego-only results on DAIR-V2X and OPV2V, across different encoder pairs and fusion strategies. Code: https://github.com/sidiangongyuan/BOLT.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Cooperative Object Detection on OPV2V (test)](<https://www.sota2.com/research/sota/cooperative-object-detection-on-opv2v-test.md>): Cooperative Object Detection; OPV2V (test); AP@30: 82.2.

- [3D Object Detection on DAIR-V2X PP→SECOND](<https://www.sota2.com/research/sota/3d-object-detection-on-dair-v2x-pp-second.md>): 3D Object Detection; DAIR-V2X PP→SECOND; AP@30: 72.6.

- [Cooperative Perception on DAIR-V2X (AP Metrics)](<https://www.sota2.com/research/sota/cooperative-perception-on-dair-v2x-ap-metrics.md>): Cooperative Perception; DAIR-V2X; AP@30: 72.9.

- [Cooperative Perception on OPV2V (AP Metrics)](<https://www.sota2.com/research/sota/cooperative-perception-on-opv2v-ap-metrics.md>): Cooperative Perception; OPV2V; AP@30: 86.8.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2605.00405>)

[PDF](<https://arxiv.org/pdf/2605.00405.pdf>)