| Underwater docking technology is a technology through which the docking station recharges the Autonomous underwater vehicle(AUV)and transmits information to and from the AUV,so as to improve the endurance of AUV.The key of underwater docking technology is to accurately navigate the AUV during the docking process.Based on the study of the docking navigation schemes at home and abroad,this paper deeply studies the multi-sensor fusion localization technology for the middle-terminal docking of AUV.The existing connection navigation mainly adopts the combination navigation mode of ultra-short baseline(USBL)and optical guidance.USBL navigation is used in the remote homing stage.After entering the optical guidance area,optical guidance navigation is used in the terminal docking stage.The shortcoming of this approach is that the update frequency of USBL positioning data is low,and there are outliers in positioning data,which may cause AUV to deviate from the optical guidance region.Especially when the sea surface light is strong,the terminal optical guidance area shrinks,which makes it more difficult for AUV to enter the optical guidance region.The existing monocular vision odometer coupled with the AUV dynamic model to improve the positioning accuracy in the transition stage between homing stage and docking stage is easily affected by underwater ocean currents and light changes.In order to solve these problems,this paper proposes a different location method based on fusion of underwater monocular vision odometer,USBL,depth sensor and attitude and heading reference system(AHRS),which does not rely on the AUV dynamic model to solve the scale of monocular vision odometer.In this solution,this paper first proposes an underwater monocular vision Simultaneous Localization and Mapping(SLAM)technology suitable for docking,which contains a tracking previous frame method based on bag of words,an addtional mismatching step and a redundant multi-layer tracking framework,which improves the operation ability of monocular vision SLAM in the underwater environment with ocean current and light variation.Then,a fusion location method of monocular vision odometer,USBL,and other sensors based on nonlinear optimization is proposed,which optimizes the AUV position and the scale of monocular vision odometer at the same time,which not only improves the accuracy and update frequency of positioning,but also solves the scale problem of monocular vision odometer.Finally,a small general AUV platform is designed and built.The docking test in swimming pool is carried out.The swimming pool test results show that the success rate of AUV docking and the terminal deviation from the docking station of the proposed scheme are better than the existing schemes. |
