Font Size: a A A

The Research To The Mobile Robots Navigation System Based On SLAM

Posted on:2015-01-10Degree:MasterType:Thesis
Country:ChinaCandidate:S XuFull Text:PDF
GTID:2308330452457165Subject:Control Engineering
Abstract/Summary:PDF Full Text Request
With the development of artificial intelligence technology, the autonomous mobilerobot is more and more widely used in various aspects. To achieve self-positioning andnavigation in a variety of complex environments is the basic for autonomous mobilerobot to complete the task, while the research of simultaneous localization and mapping(SLAM) is the key to improve the robot’s autonomy. The SLAM based on scanning laserradar is an intensive research topic in the field of mobile robotics navigation.This thesis researches the SLAM algorithm deeply based on laser radar with mobilerobot as the carrier, achieving that the mobile robot could localize and navigateautonomously. The theory and structure of SLAM is analyzed and described, then themobile robot kinematics model and the observation model based on point feature and linefeature is established. The Extended Kalman Filter algorithm is researched to implementSLAM positioning and mapping. The procedure includes following steps, predicting thestate of the system according to the robot kinematics model, using the laser data toextract environment feature and data association, using Kalman Filter to update robotposition and global map. Through the simulation, the robot can accurately estimate therobot position and landmark position by EKF-SLAM. On the basic of EKF-SLAM, theimproved algorithm FastSLAM is studied and tested. In FastSLAM, the particle filterbased robot position estimation and EKF based landmark estimation are independent ofeach other. The implement procedure of FastSLAM includes the new position sampling,new state prediction, environment map updating, particle weight calculation and particleresampling. In the simulation environment, the mobile robot could observe the landmarkaccurately and follow any given path. The FastSLAM algorithm could ensure theprecision of localizing and greatly decline the algorithm complexity of the system.As the global path planning is the basic of robot navigation, the reverse D* algorithm based on A*is proposed. In an unknown environment, given a start and a goalpoint, the mobile robot could carry out a reasonable path through the inverse D*algorithm autonomous and avoid the obstacle in real time. Finally, the main work issummarized, and several further research direction of SLAM robot navigation isproposed.
Keywords/Search Tags:Autonomous Navigation, Extended Kalman Filter, Particle Filter, DataAssociation, Global Path Planning
PDF Full Text Request
Related items