A Stereo Visual Navigation Method for Docking Autonomous Underwater Vehicles

Shuo Xu,Yanqing Jiang,Ye Li,Bo Wang,Tianqi Xie,Shuchang Li,Haodong Qi,Ao Li,Jian Cao
DOI: https://doi.org/10.1002/rob.22269
IF: 6.385
2024-01-01
Journal of Field Robotics
Abstract:Recovering and recharging autonomous underwater vehicles (AUVs) on a regular basis allows for long-term underwater activities. This research provides a vision-based navigation strategy for AUVs to independently identify and reconstruct the docking station (DS). The proposed framework includes a light beacon detection approach, a filter-based light beacon matching method, and a fusion pose estimation method for DS positioning. Four green LED light beacons are mounted symmetrically on the docking ring, enabling the stereo camera to observe them from close range. A method for detecting light beacons is proposed that assures detection accuracy by identifying false positives. On a single frame from a stereo camera, filter-based matching follows the light beacons precisely. In addition, we construct a position estimate method that significantly improves accuracy and efficiency by consisting of analyzation and iteration. A series of virtual and real-world experiments demonstrate that our methodology can provide AUVs with reliable docking navigation.
What problem does this paper attempt to address?