Research On AGV Navigation Algorithm Based On Laser SLAM

In the aviation industry,it is often necessary to test the shape and position errors of barrel components such as aircraft fuselage,in order to ensure the processing quality requirements of aircraft fuselage.At present,the laser radar is usually used to measure the aircraft fuselage,and then the measured point cloud is analyzed and tested.However,due to the overall size of the aircraft fuselage,the laser radar can not be completely measured in a single station,and station transfer operation is needed.Traditional station transfer operation usually establish a measuring field around the object to be detected.A guide rail is arranged around the measuring field,and the robot with a laser radar moves on the guide rail to achieve station-moving measurement.However,this measurement method has some disadvantages,such as the measurement field cannot be transferred,the measurement is not flexible,and the space requirement is large.Therefore,for the detection requirements of aircraft fuselage shape,the application of mobile measurement robot solution is born.Aiming at the solution of mobile measuring robot,this dissertation studies the methods of mobile chassis mapping,relocation and path planning.The main research contents and innovative achievements are as follows:(1)In precision of odometry,wheel odometry will produce the cumulative error problem in the process of movement,this paper proposes a laser odometry locating method based on matching between frames,this method takes the wheel odometry,IMU(inertial measurement unit)stance is given as initial value,and then use the ICP(iterative closest point algorithm)based on variation method for matching between frames,Finally,a laser tracker is introduced to verify that the positioning accuracy of the laser odometer proposed in this paper is higher than that of the wheel odometer,and the trajectory accuracy is improved by about 75%.(2)In terms of the mapping algorithm framework,an improved Cartographer mapping framework was proposed for the mapping function of mobile platform.It added an environmental richness discriminant module into the original framework,which was able to evaluate the current environment richness in real time and dynamically adjust input data sources.The results were verified by experiments.The final map accuracy based on the improved framework can reach 2cm/pixel.(3)Relocation algorithm in robot,this paper puts forward a kind of correlation matching method based on laser frame and map,compared with using traditional AMCL adaptive monte carlo localization algorithm,this method relocation is faster,and issue a more robust processing robot,through experiment,this method than AMCL algorithm shorten the initial relocation an aerage of 2.7 times,Relocation recovery time after abduction is reduced by an average of four times.(4)In the planning of the mobile platform path,using A * algorithm for global path planning,analysis and comparison of several heuristic function of the efficiency of path planning,Manhattan distance is chosen as the final inspiration function,validated the heuristic function of path planning efficiency than traditional Dijkstra algorithm increases by about 58% in the small map,increases by about 33% in the large map,The time elastic band algorithm(TEB)is used to carry out local path planning.The experimental results show that the vehicle can avoid dynamic obstacles flexibly.(5)In terms of navigation and positioning accuracy,the Turtlebot3 experimental platform was used to carry out positioning and navigation experiments based on laser SLAM and navigation and positioning accuracy measurement experiments.Finally,through experimental verification,the repeated positioning accuracy of the mobile platform was about2 cm,which met the design requirements.