Research On Obstacle Avoidance Planning Algorithm For Unmanned Surface Vessel

With the improvement of residents' living quality,the pollution of water surface becomes increasingly serious.At present,the method of artificial salvage is often used to clean up the floating objects on the water surface in China,which has the problems of low efficiency and high risk coefficient.It is urgent to design an environmental friendly and intelligent surface cleaning vessel.As the core problem in the field of unmanned surface vessel research,the level of autonomous obstacle avoidance reflects the intelligence of unmanned surface vessel to a certain extent.Based on the self-built unmanned surface cleaning vessel platform,this thesis studies the autonomous obstacle avoidance planning technology of unmanned surface vessel.Aiming at the problem of cleaning floating objects in small enclosed waters,this project designed and built the platform of unmanned surface cleaning vessel.The platform adopts a catamaran structure and uses the principle of differential steering to achieve small radius steering.It has high maneuverability to work in a narrow area and is equipped with the corresponding floating object salvage device.Finally,the stability and practicability of the platform are verified by the analysis of mathematical model and motion control.In order to realize the obstacle avoidance planning of unmanned surface vessel,it is necessary to use various detection equipment to obtain the surrounding environment information,and the accuracy of this information determines the final actual obstacle avoidance performance.To solve the problem that it is difficult to collect the ranging data completely in a single sensor system,this thesis selects the combination of ultrasonic sensor and lidar to perceive the surrounding environment of the unmanned surface vessel,and proposes a multi-sensor information fusion algorithm based on the combination of weighted average method and least square method.The experimental results show that the proposed method can effectively improve the accuracy and stability of the ranging system,and the error is controlled within 2cm.In view of the dynamic changes of the surrounding environment and the inability to accurately establish the kinematics model during the navigation of unmanned surface vessel,this subject based on the fuzzy logic algorithm and real-time sensor information,drawing on the driving experience of human experts and obtaining the planning information by querying the knowledge base to avoid the obstacles.At the same time,in order to solve the problem that the unmanned surface vessel needs to avoid obstacles many times when sailing in complex waters,this thesis improves the conventional fuzzy obstacle avoidance algorithm by using speed as the feedback information.Then the fuzzy neural network obstacle avoidance planning controller is designed with neural network algorithm,and the nemtwork paraeters are systematically adjusted to improve the autonomy and stability of the obstacle avoidance planning for unmanned surface vessel.Finally,experiments are carried out on the simulated experimental platform and the actual water environment to test the obstacle avoidance performance of unmanned surface vessel.The experimental results show that compared with the traditional fuzzy obstacle avoidance planning algorithm,the time and distance taken by this method in continuous obstacle environment are reduced by more than 15%,and the success rate of obstacle avoidance in complex environment is more than 90%.Therefore,the proposed method can basically meet the real-time obstacle avoidance requirements of unmanned surface cleaning vessel platform in small enclosed waters.