Simultaneous Localization And Mapping For Guide Robot Based On Multisensor

Eyes are the most important tool for human to aware this world,blindness and visual impairment seriously affect the quality of human daily life.Blindness and visual impairment may be completely solved in the future,but in terms of the present stage,it is necessary to research the blind guiding subject.Generally,most of the blind and visual impairment use walking sticks for blind guiding.The world is becoming more and more complex with the rapid social development,walking stick cannot satisfy the blind guiding task,however,the biological blind guiding like seeing-eye dogs can't be popularized because the high cost and long period of training and the short lifetime.The beat way to solve the problem of blind person is to research and develop high-performance blind guiding robot,due to the intelligentize machine becomes more and more mature.In this paper,the simultaneous localization and mapping technology will be uesd on blind guiding robot,and different maps will be built with multiple sensors,then path planning and obstacle avoidance will be used based on the map.In complex environment,the multi-sensor data fusion will be used to add the information in the map,in order to accurately complete the complex blind guiding mission.The main content is as follows:1.Two types of SLAM algorithm are analyzed,one is filter SLAM which is based on the laser scan matching,the other is graph optimization SLAM which is based on the image processing and point cloud registration.In filter SLAM,Extended Kalman Filter and Particle Filter are compared and simulated.In graph optimization SLAM,I explain feature point extraction and matching,motion estimation and optimizing,pose graph building,loop closure,pose graph optimization,and 3D map building.2.The data from LIDAR and Kinet will be fused to deal with the complex environment.The blind guiding robot can't get enough information with single sensor.In order to complete the complex blind guiding mission,the blind guiding robot will carry LIDAR and Kinet at the same time.Two methods of data fusion are presented in this paper,they are map level and data level,which can get more accurate map.3.Hybrid route planning algorithm is presented.Because of the complex and changeful environment and the high-precision navigation,The blind guiding robot need a hybrid route planning algorithm to complete global path planning and dynamic obstacle avoidance.The algorithm mixes a global path planning using A* algorithm and a local path planning using improved artifical potential field.4.An experimental platform is set up to conduct experiments.Building 2D maps in small room and big environment with LIDAR using filter SLAM algorithm.Building 3D maps with Kinect using graph optimization SLAM algorithm.Testing the data fusion algorithms in aspecific environment,fusing the maps in map level and data level.Testing the hybrid route planning algorithm.