A High-accuracy Event-based Underwater SLAM System

2026-06-17

AI Summary

This work addresses the limitations of existing time-surface (TS)-based event camera underwater SLAM, which often fails under conditions of high camera motion, large-baseline stereo configurations, and scenes with repetitive textures. We present the first high-precision, event-driven underwater stereo SLAM system, introducing a structure tensor– and gradient-guided metric to assess TS information density. A two-stage TS optimization strategy is devised: a prior TS is predicted via Bayesian optimization before initialization, and asynchronous online local search dynamically refines it during tracking. Data association is enhanced through prior disparity constraints, while a “most recent observation first” triangulation scheme improves robustness. Evaluated on our newly collected real-world underwater event dataset (UWE) and public benchmarks, the proposed system significantly outperforms existing methods in complex underwater environments. Code and data will be made publicly available.

Abstract

While event cameras offer immense potential for underwater SLAM, existing Time Surface (TS)-based methods prove highly unreliable when deployed underwater. Fluctuating camera velocities severely degrade TS imaging quality, while wide stereo baselines and repetitive underwater textures induce critical matching failures, frequently triggering system failure. To overcome these challenges, we develop the first high-accuracy event-based underwater stereo SLAM system. A structure-aware metric for TS is designed based on structure tensor coherence and gradients to quantitatively evaluate TS structural information density. By decoupling the optimal TS generation into two distinct stages based on system initialization, Bayesian Optimization(BO) first predicts an optimal prior TS sequentially before initialization while we set an asynchronous online local searching method periodically to obtain appropriate TS in real-time during the tracking stage. We use the prior disparity to guarantee precise data association and "latest-observation-first'' triangulation mechanism to realize stable triangulation. As a benchmark for these solutions and a resource for the community, we also contribute UWE, the first high-quality real-world underwater event dataset containing variable camera motions, complex textures and different trajectory features. Extensive evaluations on public datasets and UWE show the competitive accuracy performance of the proposed SLAM system compared to the state-of-the-art event-based method. The code and data will be open-sourced.

Problem

• underwater SLAM
• event camera
• Time Surface
• matching failure
• stereo baseline

Innovation

• event-based SLAM
• underwater robotics
• Time Surface optimization
• structure-aware metric
• Bayesian Optimization