Vision-based Underwater Environment Modeling And Trajectory Planning

Abundant marine resources are of great significance for the development and utilization of marine resources.The equipment currently used for marine development and exploration is mainly underwater vehicles.When an underwater vehicle is sailing in the ocean,it will inevitably be affected by the complex and changing marine environment.Traditional environmental models usually only consider information about known marine terrain,and cannot fully reflect the true situation of the marine environment.The information available for path planning is incomplete.Therefore,in order to better assist the underwater submarine in route planning,it is urgently necessary to describe the marine environment information completely and accurately.This thesis focuses on the vision-based underwater environment modeling and trajectory planning.Through the collection and processing of underwater environment information using a visual acquisition system to establish an accurate and complete underwater 3D model,and finally rely on a set of optimization algorithms to achieve underwater positioning and trajectory planning.First,this thesis designs an underwater robot system as an experimental platform.Its hardware system is mainly composed of Nvidia TX2 and pixhawk flight control.The former is responsible for processing video data signals and running some medium-sized calculation programs.The latter is mainly responsible for robot motion control.The entire software system is based on the ROS platform and uses a distributed communication framework.Secondly,for the underwater environment modeling problem,this paper adopts the method of combining ORB-SLAM and OCTOMAP to establish a dense three-dimensional environment.By studying the representation form of map data,the octree grid data is selected as the map representation format.After processing the depth image data collected by the Kinect depth camera,it is converted into three-dimensional octomap occupancy data.The ORB-SLAM algorithm is used to provide the pose of the camera,which is combined with the above three-dimensional occupancy grid data to create a dense underwater three-dimensional environment map.Through camera calibration and octomap simulation experiments,it can be seen that the modeling effect of this method can reach expectations.Finally,it addresses the problem of path planning for underwater robots.This paper designs a trajectory and geometric path tracking controller based on the PID algorithm.By studying the dynamic model equations of the underwater robot,the model parameters obtained.The controller simulation experiments and analysis are performed using Matlab.It can be seen from the results of Matlab simulation experiments that the controller designed is very clear.