Research On USV's Environmental-perception Based On Multi-information Fusion

In the complex and changeable marine environment,real-time and accurate multi information fast fusion of environmental perception is the premise of autonomous navigation and the basis of collision avoidance decision-making control.In order to solve the problem of real-time perception and unified description of multi information of unmanned surface craft in unknown dynamic marine environment,the theory of multi information fast fusion and optimization of onboard navigation radar,electronic chart and integrated navigation is studied.In this paper,taking the blue signal high-speed USV of Dalian Maritime University as the research object,a multi information fusion environment sensing system for USV is designed.Taking the inertial navigation system as the motion state reference information,the radar image is superimposed on the electronic chart for the unified scale description.At the same time,the fast feature extraction method is used to calculate the motion state information of obstacles,and the environmental information is analyzed.The sensing accuracy is optimized.The main work is as follows:1.In order to solve the problem of matching and superimposing electronic chart and radar image,a contour matching algorithm is selected.First of all,real-time sea area information is acquired by shipborne radar to generate radar image,and inertial navigation information is received by serial port to analyze the position information and course information of USV.Then,the least squares is used to establish the relationship between the coordinates and the pixel points.By solving the inertial navigation information,the position information is solved into the electronic chart.The image processing technology is used to adjust the size and orientation of the radar image,process the radar image and the electronic chart,and achieve the first step of matching and fusion.Finally,the Hausdorff distance matching algorithm is used to correct the matching points through the similar comparison of the contour,so as to realize the accurate matching and superposition of radar image and electronic chart.2,In order to solve the problem that the radar image can not give the obstacle information,an attitude algorithm is designed.First,the contour area and position are calculated by extracting the contour of obstacles.Then,comparing the position changes of obstacles in two consecutive images,we design the method of calculating speed,determine the direction of 0 degree heading,and use the information of speed to design the method of calculating heading to solve the single object dynamic obstacle information.Finally,aiming at the multi-target obstacle information extraction,through the change of the obstacle position,the multi-target accurate corresponding method is designed.In order to achieve one-to-one correspondence of multiple targets under different images,the speed and heading of obstacles are solved.3.To verify the effectiveness of the method,design a multi-information fusion USV environment perception system and build a hardware experiment platform.First,the USV is used as the experimental platform and the shipborne radar,shipborne inertial navigation system,and electronic chart development system are used as multiple information sources.Using VS2013 and QT5.7 as the software development environment,build an USV environment perception system Then,design software algorithm programming,realize the division of labor for multiple information,and integrate the processed information.Finally,verify the effectiveness of the algorithm through experiments.