Research On Deep-sea AUV Real-time Collision Avoidance Method

Currently,with the increasing research about the development of oceans,people put forward higher requirements to the use of Autonomous Underwater Vehicle(AUV).AUV is hoped to complete tasks,such as seabed topography tracking,seabed mineral resources exploration photographs and sailing.these tasks give high requirements to AUV,which need AUV is not more than 5m away from the seabed,and the speed is less than two.In addition,as the unknown seabed topography and complex,we can not plan the AUV path advanced,which requires the AUV must be carried out in real-time collision avoidance.Most abroad AUV are designed for specific tasks,the number of AUV which has a good real-time tracking and avoid collision,forcing on the high submarine navigation and seabed topography is still relatively small.Deep seabed topography terrain is as complex as land,while deep-sea underwater robot means very limited perception of the environment,altimeter and depth gauge are present more background noise and large measurement error.Furthermore,since the deep-sea underwater robot and sonar ranging carry forward looking sonar detection range is very limited,usually about 100 meters,then consider each sonar detection angle limit,which means able to perceive the deep underwater environmental information is incomplete,and very limited.How to choose a filtering algorithm to estimate the slope of the terrain accurately and desigen the better robustness of real-time collision avoidance method on deep sea AUV real-time obstacle avoidance is difficult.The main contents and conclusions of this study are summarized as follows:1)According to altimeter and depth gauge with both background noise and large measurement error,this paper proposes estimation based on extended Kalman filter and least squares undersea terrain slope,and algorithm simulation.2)For AUV's task of seabed terrain tracking and navigating height,this paper presents AUV vertical motion control based on sliding mode control,and establishs AUV kinematics and dynamics model,with the algorithm simulation.3)With the limit of capacity of the deep sea AUV sensors to detect environmental perception,this paper is proposed AUV real-time collision avoidance methods based onnear domain diagram navigation,and algorithm is verified by half-physical simulation platform.4)semi-physical simulation and test verification lake test are taken on the proposed deep-sea underwater robot in real-time collision avoidance methods.Autonomous Underwater Vehicle real-time collision avoidance is an extremely challenging research topic,this paper is only for a kind of underactuated AUV,based on extended Kalman filter,sliding mode control and navigation domain diagram to design a near real-time collision avoidance system implementation.Then test the feasibility and effectiveness of the system with simulation and experimental verification.