Research On Autonomous Obstacle Avoidance Of Unmanned Surface Vessel Based On Multi-sensor Fusion SLAM

Unmanned surface vessel(USV)is a kind of task-based water surface motion platform with multiple sensors.It has the characteristics of intelligent autonomy,flexibility,safety and concealment,and plays an important role in military and civilian fields.In the process of operation,unmanned vessels will inevitably encounter obstacles such as buoys,ships and piers.In order to avoid collision accidents,USV must detect obstacles in the working waters and take safe paths to avoid obstacles.In this thesis,simultaneous localization and mapping(SLAM)technology is introduced into USV.At the same time,with the help of multi-sensor fusion technology,USV can complete mapping and autonomous obstacle avoidance in real water environment.The main research contents and work are as follows:Firstly,through reading the literature at home and abroad,the research status of multisensor fusion technology,SLAM technology and USV obstacle avoidance is analyzed,and the basic research scheme of the subject is proposed.The multi-sensor fusion SLAM and the required ROS technology are analyzed.The models and working principles of LIDAR,IMU and GPS sensors are derived,and three map models commonly used by mobile robots are analyzed.Secondly,the multi-sensor fusion unmanned surface vessel positioning and mapping are analyzed and tested.The traditional USV positioning uses GPS,which will fail due to factors such as trees and poor satellite signals.This thesis uses a fusion method of GPS and IMU loose combination for positioning.After completing the fusion positioning of the USV,the GPS and IMU fusion data are combined with the LIDAR data to construct the water area map.The map construction algorithm uses the improved GMapping algorithm.The improved GMapping algorithm optimizes the particle resampling step on the basis of the traditional algorithm.By setting two weight thresholds,the particles are divided into low weight,medium weight and high weight particle regions.By resampling the low-weight set particles and copying the high-weight set particles,the diversity of particles in the mapping process is improved.The experiment verifies that the mapping accuracy of the improved algorithm is improved.The global path planning algorithm and the local obstacle avoidance algorithm are simulated and analyzed on the grid map.Through simulation comparison,the A * algorithm is selected as the global path planning,and the DWA algorithm is used as the local obstacle avoidance.And the two algorithms are combined with the USV obstacle avoidance scheme.Finally,this thesis builds the USV platform,and carries out the water surface mapping and obstacle avoidance test of the USV.The hardware system of USV adopts a three-layer structure based on data layer,computing layer and control layer.The data layer is responsible for carrying relevant sensors to obtain environmental data.The computing layer receives data from the control layer and the data layer for processing,and sends control commands to the control layer.The control layer is responsible for realizing the motion control of the USV and feeding it back to the computing layer.The designed USV platform is used for obstacle avoidance experiment.It can be seen from the experimental results that the unmanned surface vessel can safely avoid obstacles in the water environment,and the navigation process is also relatively stable.The real ship experiments to verify the effectiveness of the autonomous obstacle avoidance scheme for the multi-sensor fusion SLAM proposed in this thesis,and the unmanned surface vessel can complete the autonomous obstacle avoidance navigation in the waters.