Research On The Docking Guidance Algorithm Of Autonomous Underwater Robot Based On Vision

Autonomous docking is of great significance for autonomous underwater robot.Docking with the submarine base means that AUV can realize resident underwater operation,docking with the submarine can play important role in crisis rescue,docking with the recovery unit can realize AUV underwater autonomous recovery and so on.The same as the autonomous air refueling of aircrafts and the rendezvous and docking of Shenzhou 9 with Tiangong 1,AUV autonomous docking is the process of AUV automatically detecting the target,getting close to the target and finally enter the recovery unit.This paper focus on the problem of close optical guidance during the process of AUV docking.The portable AUV developed by SCMITE Shenyang Institute of Automation,Chinese Academy of Sciences,is adopted to carry out experiments.Based on the hardware and software performance of the AUV,proposed a vision localization algorithm.The visual sensor plays an important role as an important sensor for AUV intelligence.Due to the special underwater environment,the development of AUV vision slowly,most of the application only stay in close range image acquisition.The use of visual docking is an intelligent application of AUV.According to the feature that the effective distance of underwater image is short,a guidance light source is equipped on the recovery unit,and an independent high resolution visual gathering and processing system is developed.The communication between the visual processing system and the AUV is realized through MOOS.In view of the characteristics of underwater image noise,morphological filtering is adopted and good results have been obtained.Using the MATLAB calibration toolbox to calibrate the camera which is encapsulated in the AUV head segment,the reliable camera distortion parameter and internal parameter matrix are obtained.In this paper,the problem of extracting difficulty of eight guiding lamps for underwater image is presented,and the feature point extraction is proposed by the self-tuning threshold segmentation algorithm.According to eight guiding lights matching problem,put forward based on pixel coordinates normalized feature point matching method,complete guiding light positioning,and calculate the center of the plane guiding lights relationship with the relative position of the camera.Finally,the design of pool docking experiment is designed to verify the reliability of the algorithm.Research on autonomous underwater visual docking guidance algorithm,this algorithm is to realize the AUV underwater autonomous docking indispensable software modules,is of great significance for the future for the connection of the AUV underwater autonomous and practical value.