# LiDAR Teach, Radar Repeat: Robust Cross-Modal Navigation in Degenerate and Varying Environments

Source: [LiDAR Teach, Radar Repeat: Robust Cross-Modal Navigation in Degenerate and Varying Environments](<https://www.sota2.com/research/paper/69f979d8c0f56ed843089a83>)

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

Data updated: 2026-05-06T05:00:18.259Z

Authors: Renxiang Xiao, Yichen Chen, Yuanfan Zhang, Qianyi Shao, Yushuai Chen, Yuxuan Han, Yunjiang Lou, Liang Hu

Published: 2026-05-04T16:46:54.000Z

## Abstract

Long-term autonomy requires robust navigation in environments subject to dynamic and static changes, as well as adverse weather conditions. Teach-and-Repeat (T\\&R) navigation offers a reliable and cost-effective solution by avoiding the need for consistent global mapping; however, existing T\\&R systems lack a systematic solution to tackle various environmental variations such as weather degradation, ephemeral dynamics, and structural changes. This work proposes LTR$^2$, the first cross-modal, cross-platform LiDAR-Teach-and-Radar-Repeat system that systematically addresses these challenges. LTR$^2$ leverages LiDAR during the teaching phase to capture precise structural information under normal conditions and utilizes 4D millimeter-wave radar during the repeating phase for robust operation under environmental degradations. To align sparse and noisy forward-looking 4D radar with dense and accurate omnidirectional 3D LiDAR data, we introduce a Cross-Modal Registration (CMR) network that jointly exploits Doppler-based motion priors and the physical laws governing LiDAR intensity and radar power density. Furthermore, we propose an adaptive fine-tuning strategy that incrementally updates the CMR network based on localization errors, enabling long-term adaptability to static environmental changes without ground-truth labels. We demonstrate that the proposed CMR network achieves state-of-the-art cross-modal registration performance on the open-access dataset. Then we validate LTR$^2$ across three robot platforms over a large-scale, long-term deployment (40+ km over 6 months), including challenging conditions such as nighttime smoke. Experimental results and ablation studies demonstrate centimeter-level accuracy and strong robustness against diverse environmental disturbances, significantly outperforming existing approaches.

## 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.

- [Cross-modal Registration on VoD sequence 01](<https://www.sota2.com/research/sota/cross-modal-registration-on-vod-sequence-01.md>): Cross-modal registration; VoD sequence 01; RYE: 0.219.

- [Cross-modal Registration on VoD sequence 02](<https://www.sota2.com/research/sota/cross-modal-registration-on-vod-sequence-02.md>): Cross-modal registration; VoD sequence 02; RYE: 0.221.

- [Cross-modal Registration on VoD sequence 03](<https://www.sota2.com/research/sota/cross-modal-registration-on-vod-sequence-03.md>): Cross-modal registration; VoD sequence 03; RYE: 0.217.

- [Cross-modal Registration on VoD sequence 04](<https://www.sota2.com/research/sota/cross-modal-registration-on-vod-sequence-04.md>): Cross-modal registration; VoD sequence 04; RYE: 0.213.

- [Cross-modal Registration on VoD sequence 14](<https://www.sota2.com/research/sota/cross-modal-registration-on-vod-sequence-14.md>): Cross-modal registration; VoD sequence 14; RYE: 0.16.

- [Cross-modal Registration on VoD sequence 19](<https://www.sota2.com/research/sota/cross-modal-registration-on-vod-sequence-19.md>): Cross-modal registration; VoD sequence 19; RYE: 0.191.

- [Open-loop Registration on VoD sequence 01](<https://www.sota2.com/research/sota/open-loop-registration-on-vod-sequence-01.md>): Open-loop Registration; VoD sequence 01; RTE (m): 0.092.

- [Open-loop Registration on VoD sequence 02](<https://www.sota2.com/research/sota/open-loop-registration-on-vod-sequence-02.md>): Open-loop Registration; VoD sequence 02; RTE (m): 0.106.

- [Open-loop Registration on VoD sequence 03](<https://www.sota2.com/research/sota/open-loop-registration-on-vod-sequence-03.md>): Open-loop Registration; VoD sequence 03; RTE (m): 0.1.

- [Open-loop Registration on VoD sequence 04](<https://www.sota2.com/research/sota/open-loop-registration-on-vod-sequence-04.md>): Open-loop Registration; VoD sequence 04; RTE (m): 0.066.

- [Open-loop Registration on VoD sequence 14](<https://www.sota2.com/research/sota/open-loop-registration-on-vod-sequence-14.md>): Open-loop Registration; VoD sequence 14; RTE (m): 0.116.

- [Open-loop Registration on VoD sequence 19](<https://www.sota2.com/research/sota/open-loop-registration-on-vod-sequence-19.md>): Open-loop Registration; VoD sequence 19; RTE (m): 0.121.

- [Closed-loop Navigation on Trajectory E](<https://www.sota2.com/research/sota/closed-loop-navigation-on-trajectory-e.md>): Closed-loop Navigation; Trajectory E; Yaw Error Average (YAE): 2.151.

- [Closed-loop Navigation on Trajectory E (ATE)](<https://www.sota2.com/research/sota/closed-loop-navigation-on-trajectory-e-ate.md>): Closed-loop Navigation; Trajectory E; ATE: 0.067.

- [Closed-loop Navigation on Trajectory C](<https://www.sota2.com/research/sota/closed-loop-navigation-on-trajectory-c.md>): Closed-loop Navigation; Trajectory C; YAE: 2.468.

- [Closed-loop Navigation on Trajectory D](<https://www.sota2.com/research/sota/closed-loop-navigation-on-trajectory-d.md>): Closed-loop Navigation; Trajectory D; Yaw Error Average (YAE): 2.326.

- [Closed-loop Navigation on Trajectory F](<https://www.sota2.com/research/sota/closed-loop-navigation-on-trajectory-f.md>): Closed-loop Navigation; Trajectory F; Yaw Error Average (YAE): 1.982.

- [Closed-loop Navigation on Trajectory C (ATE)](<https://www.sota2.com/research/sota/closed-loop-navigation-on-trajectory-c-ate.md>): Closed-loop Navigation; Trajectory C; ATE: 0.08.

- [Closed-loop Navigation on Trajectory D (ATE)](<https://www.sota2.com/research/sota/closed-loop-navigation-on-trajectory-d-ate.md>): Closed-loop Navigation; Trajectory D; ATE: 0.067.

- [Closed-loop Navigation on Trajectory F (ATE)](<https://www.sota2.com/research/sota/closed-loop-navigation-on-trajectory-f-ate.md>): Closed-loop Navigation; Trajectory F; ATE: 0.069.

- [Closed-loop Navigation on Trajectory B](<https://www.sota2.com/research/sota/closed-loop-navigation-on-trajectory-b.md>): Closed-loop Navigation; Trajectory B; Yaw Error Average (YAE): 2.657.

- [Closed-loop Navigation on Trajectory G (ATE)](<https://www.sota2.com/research/sota/closed-loop-navigation-on-trajectory-g-ate.md>): Closed-loop Navigation; Trajectory G; ATE: 0.077.

- [Robot Navigation on Trajectory A](<https://www.sota2.com/research/sota/robot-navigation-on-trajectory-a.md>): Robot navigation; Trajectory A; Success Rate: 3.

- [Closed-loop Navigation on Trajectory G](<https://www.sota2.com/research/sota/closed-loop-navigation-on-trajectory-g.md>): Closed-loop Navigation; Trajectory G; Yaw Error Average (YAE): 2.949.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2605.02809>)

[PDF](<https://arxiv.org/pdf/2605.02809.pdf>)
