Service Robot Positioning Based On Particle Filter In Indoor Dynamic Environment

With the continuous development of artificial intelligence,people’s requirements for the intelligent level of robots are gradually improved.As a new member in the field of robots,service robot in recent years has a growing potential of applications in hospitals,shopping malls,museums and other places.In these places,service robots have to deal with the change of the environment,such as walking,opening,or closing door etc.These moving obstacles will effect the localization accuracy of robot,and as a reslt they will cause the positioning failure of a mobile robot,which has a great influence on autonomous navigation of service robot.Based on this background described above,this paper focuses on the service robot mapping and self-localization by using a particle filter.The main works of this thesis are listed as below:This thesis begins with description of the service robot and analysis of the different forms of Bayesian theory,which is followed by the discussion of the particle filter theory.At the same time the probabilistic motion model and observation model are established.Then,aiming at the problem that the traditional grid-based Fast SLAM algorithm is complex and the edge of the map is poor,this thesis uses scan matching algorithm based on gradient descent to correct the cumulative error brought by the odometry in order to improve the accuracy of the robot map.At the same time,the algorithem was designed and implemented on the ROS(Robot Operating System)platform.Secondly,aiming at the problem self-localization of service robot in indoor dynamic environment,this thesis improves the traditional particle filter algorithm with a laser radar,which is used to identify moving obstacles and eliminate them so as to reduce the impact of moving obstacles on robot positioning accuracy.Based on the improved algorithm,the robot in indoor dynamic scene has higher positioning accuracy.The improved algorithm is designed and implemented on the ROS flatform.Next,this thesis designs experiments in the laboratory corridor with turtlebot2 as the experimental platform,and the experimental results show that the system can realize map creation and localization function,and the system has strong robustnessFinally,the research results revealed by this thesis are summarized and the shortcomings of the thesis are discussed.Additionally,the author makes a discussion and outlook on further developments and research direction in this field.