Research On Real-time Localization And Online Mapping Of An Autonomous Underwater Vehicle For Water Tunnel Inspection

The water tunnel is the most crucial part of the long-distance water supply project.Because of full of the water,high pressure,large velocity,long distance,dark and complex environment,cracks are the common problem after a long running.Above all the inspection of the water tunnel is a key to ensure the continuous water supply of the city.An Autonomous Underwater Vehicle(AUV)for water tunnel inspection can effectively solve the inspection problem of water tunnel and ensure the safety of reservoir project.Since the traditional positioning algorithm cannot be realized in the environment of water tunnel,this paper has carried out a series of studies on the real-time positioning and online mapping of the AUV,especially on the key algorithms and technical problems.The researches and major achievements of this paper are as follows.Firstly,introduced the particularity of tunnel environment,analyzed the reasons why the previous underwater inspection devices cannot meet the requirements of inspection task,and then studied the development status of SLAM(Simultaneous Localization and Mapping)algorithm,which is the key algorithm used in this paper.Secondly,introduced the design of the AUV,and established the motion model,observation model and characteristic model of AUV.Thirdly,the IMU(Inertial measurement unit),sonars and other sensors have been analyzed and corrected.Working mode of dynamic weighted round-robin can reduce the mutual interference of sonars.Then analyzed the error model of IMU.According to the practical environment of AUV inspection in water tunnel,a method is designed which can effectively restrain the accumulated angle error of IMUFourthly,in view of the special environment of water tunnel,a method of real-time positioning and online mapping of the AUV is proposed.According to the fact that the pressure in the water tunnel will increase when the AUV is working.With the operation of the AUV,a special positioning method for tunnel inspection is designed when the forward sonar cannot observe the landmarks,and this method can achieve positioning and mapping better accuracy.When the sonar can observe landmarks,the AUV will perform the improved EKF-SLAM algorithm.The landmark extraction and data association are improved with practical situation.In terms of error correction,every work interval,the measured data from ranging sonars in both sides of AUV can correct the horizontal position error,this method greatly improved the accuracy of positioning and mapping.Finally,through the algorithm simulation of real-time positioning and online mapping of the collected water tunnel experimental data,the results show that the accuracy of positioning and mapping is very high.The improved EKF-SLAM algorithm based on the practical environment of water tunnel in this paper can effectively restrain the increase of positioning error.According to the simulation results,a special positioning method for tunnel inspection also achieved better mapping accuracy.