Research On The Problem Of Simultaneous Localization And Mapping For Multiple Mobile Robot Based On Kinect

Mobile robot 3D SLAM refers to that the mobile robot builds the 3D environment maps with equipped sensors,and estimate their trajectory simultaneously.It is the key issue to achieve autonomous mobile robot,it's also a difficult problem in the field of robotics.Compared to single-robot 3D SLAM,multiple robot cooperative simultaneous localization and mapping not only improve the efficiency of the composition,but also increases the robustness of the whole system.The difficulties of multiple robot 3D SLAM mainly include the communication between robots and their local map fusion,besides the difficulties in single-robot 3D SLAM.State estimation,data association and map fusion problem related to multiple robot simultaneous localization and mapping are studied in this paper.The main works of this thesis are as follows:(1)Firstly,the problem of single-robot 3D SLAM was studied.Through reading sensor data,visual odometry,loop closure detection and back-end optimization,a globally consistent 3D environment map and robots' trajectory can be achieved.In feature detection and matching,a data dimension reduction algorithm called Principle Component Analysis(PCA)is proposed.It can improve the efficiency of single-robot 3D SLAM.In local map construction,a key frame selection algorithm is proposed to improve the accuracy and robustness of building 3D map.(2)Secondly,the problem of multiple robot 3D SLAM was studied.The overview construction of the multi-robot system and wireless communication technology were introduced.Homogeneous coordinate transformation is used to solve the map align problem.According to whether two robots meet or not,two map fusion approach are approved.In the first approach,relatively observed by two robots to get the overlaps of two sub-map in order to map fusion.In the second approach,the multiple map fusion problem is transformed into objective optimization problem,and the best transformation matrix between two sub-map is found to realize map fusion.(3)Two approaches are used on the proper situation so that multiple robot 3D SLAM is achieved.(4)Finally,the proposed improved algorithms,such as a data dimension reduction algorithm PCA and key frame selection algorithm,are applied in multiple robot 3D SLAM.Experiments show that the improved algorithms work well on multiple robot simultaneous localization and mapping.At the end of the paper,some conclusions of the research work are made and some outlook of the next research direction or problems are discussed briefly.