Obstacle Avoidance And Target Tracking For Underwater Robot Based On Sonar And Vision

The underwater robot is an indispensable tool to replace the human being under the sea.The application of obstacle avoidance and target tracking of underwater robot is very extensive.It plays an important role in the care of diver's safety,marine biological research,channel construction and safe defense.However,the underwater environment is complex and special.It is difficult to collect clear sensor information,and determine the specific stress model.What's more the traditional obstacle avoidance and tracking algorithms can't be used directly under water.The current study of underwater target tracking only take advantage of sonar or camera,which due to the surrounding information perception is not comprehensive,and result in the accuracy is not satisfied in practical application In order to solve these problems,this thesis designs a set of target tracking and obstacles avoidance system based on the underwater robot,which can track mobile target and avoid obstacles.The design of the system includes the acquisition and processing information of the sensor and camera,the design of the motion controller,the design of obstacle avoidance and target tracking strategy.Firstly,it constructs communication between underwater robot and the control computer,and process information of the underwater robot sonar and camera to get richer and clearer information of the surrounding environment.Secondly,it is based on the dynamic model of underwater robot to control the robot motion by using the fuzzy PID controller.Finally,it designs the strategy of obstacle avoidance and target tracking tasks.The main contents of this paper include:1)Construction of the camera imaging model of underwater,calibration the camera and correction distortion.What's more,we propose an approach of fusing images to enhance underwater image.2)Communication system between sonar and computer.we collect the sonar echo data,map the sound image,and remove the noise of the sound image.3)Models kinematics and dynamics of the underwater robot and a method of motion control based on fuzzy PID is proposed to control the underwater robot.4)Proposes an improved Artificial Potential Field method,which can solve the local minimum problem,and excuse real time obstacle avoidance in constant current of the underwater robot by method of speed synthesis.5)Put forward an improved tracking algorithm based on kernel correlation filter,which can track the target accurately and quickly even if the underwater target is deformed or occluded.