Research On Mobile Robot Navigation Based On Kinect Sensor

At the beginning of 21th century,Mobile robot is an important component part of robots which can act as an important role to realize artificial intelligence research.Unknown environment exploration and exploitation by mobile robot has made outstanding effect on human beings.With the development of technology,the requirement for mobile robot is becoming increasingly higher in various fields,including industry,medical treatment,national defense,search and rescue etc.Mobile robot possess its‘own thought（independence,intelligence）is a difficult work to deal with in the future.The paper targets at realizing mobile robot map-building,self-localization and navigation.The experimental platform contains Kinect sensor and Kobuki base.Research on mobile robot navigation is based on SLAM,Path-planning.Mobile robot navigation steps contain:first,forehead data obtain,fusing the data collected by Kinect and odometry;then,making mobile robot accurate localization based on RBPF-SLAM,building incremental map by updated pose and the data collected by sensor;By global path planning algorithm A^*or Dijkstra creates a shortest path based on the grid-map built before;when mobile robot forwards in the shortest path,making local obstacle avoidance by Dynamic window approach algorithm.The work details of the paper are as follows:1.With the analysis of the important status and developing prospect of mobile robot,make an overview of navigation system.With the requirement of SLAM and Path-planning during unknown environment exploration,designing a solution based on ROS and Kinect sensor.2.Solved the problem about mobile robot equips with 1D or 2D sensor usually has some distance from the ground,it may cause robot mistakenly perceive the shape of obstacle（i.e.L shape）under the sensor‘s field of view,as a result of missing some importance data and mistakenly expressing the shape of obstacle,It would not be as efficient as it should to be.3.SLAM:on account of Gaussian hypothesis during the extended Kalman filter,bringing in EKF-SLAM,the step of EKF algorithm used in SLAM had been discussed,the feasibility of EKF algorithm used in SLAM had been proved.In order to overcome the shortcomings of EKF-SLAM,the paper used the RBPF-SLAM,coarse correction position and pose used as input of Rao-Blackwellized particle filter,outputting precise position and pose and building incremental environment map in the end.4.With the analysis of fundamental principles and implementation process of global path planning(A^*algorithm and Dijkstra algorithm),in the grid-map built before,generating an optimal path by A^*algorithm.Then,local path planning was made by DWA algorithm,completing the control of the mobile robot.Finally,mobile robot can make obstacle avoidance during navigation.5.Accomplished the design of Software and hardware in Turtlebot mobile robot under the ROS platform,mobile robot navigation had achieved.Reliability and feasibility of designed system had been proved by the experiment in the paper.