Font Size: a A A

Research On Localization And Mapping Algorithm Of Outdoor Cleaning Robot Based On 3D Laser

Posted on:2021-04-04Degree:MasterType:Thesis
Country:ChinaCandidate:C LiFull Text:PDF
GTID:2480306308975479Subject:Mechanical engineering
Abstract/Summary:PDF Full Text Request
The research of outdoor mobile robot localization and mapping algorithm has been the main subject in the field of autonomous mobile robot research.How to locate in complex outdoor scenes and how to create effective obstacle grid maps in scenes with many obstacles to facilitate path planning for mobile robots has always been the focus and difficulty of robotics research.In this dissertation,an outdoor cleaning robot working in areas such as parking spaces between roads is taken as the research object,and the positioning and mapping algorithms of outdoor cleaning robots based on 3D lasers are studied.By detecting obstacles to create a map suitable for outdoor cleaning scenes,combining the advantages and disadvantages of 3D lidar and GPS for accurate and fast positioning,to ensure the performance of the cleaning robot.The research in this article is divided into the following aspects:1.Research on the construction algorithm of outdoor passable area.The point cloud depth map is used to describe the 3D point cloud,and the neighborhood information is explicitly displayed to reduce the time consumption of the algorithm on the computing platform.The ground plane fitting algorithm is used to separate ground points and non-ground points,and pass the octree Trees are used to construct a local obstacle grid map;using an improved depth clustering algorithm,special obstacles with a small volume are initially detected,and special obstacles such as leaves are further detected through the laser intensity distribution of the point cloud,creating an obstacle grid In the map,a passable area is constructed to facilitate the path planning of the robot.2.Research on 3D LIDAR fusion GPS positioning algorithm.Construct a pose covariance matrix describing the pose information of the robot.Use GPS prior deviation and positioning factors to construct a GPS pose covariance matrix.Use point cloud matching to construct a laser odometer pose covariance matrix.Posture covariance matrix and other information,using GPS and Lidar sensor information to improve the pose accuracy and robustness of the robot using a pose map optimization method;adding closed-loop detection,based on the global GPS information and the result of point cloud matching as a closed loop The judgment basis for detection further improves the accuracy of the robot's posture?3.Research on lightweight laser mileage calculation method.Based on the existing point cloud processing,point cloud segmentation and other methods are used to eliminate unreliable feature points with excessive local curvature in the point cloud,reduce the number of matching feature points,and reduce the calculation time of the laser odometer part;reduce local curvature A larger number of line features improves the accuracy of matching while reducing the amount of calculation;adding plane motion constraints to the laser odometry pose calculation to improve the robot's pose accuracy and robustness on rough roads.4.Experimental research on positioning and mapping algorithm of outdoor cleaning robot based on 3D laser.Carry out system design and robot platform construction,and use simulation test and actual test to verify the feasibility and effectiveness of the algorithm.
Keywords/Search Tags:Traffic area, improved depth clustering, GPS-Lidar fusion positioning, lightweight
PDF Full Text Request
Related items