Study On Localization And Mapping Of A Mobile Robot Based On Sensors Fusion

In order to study the autonomy of the indoor mobile robot, an experimental mobile robot platform and its simulation model based on Matlab were completed. And motion control, posture analysis, and SLAM (Simultaneous Localization and Mapping) of the robot were mainly studied.The ROS operating system was chosen as the robot software development platform. These robot software algorithms included sensors fusion for posture estimation, SLAM based on particles filter and so on. The ROS provided messaging transform mechanism for software modules integration. The ARM7microcontroller was used to implement closed-loop motion control, and the AVR microcontroller handled Inertia measurement data collection. These interfaces and drivers of all modules were also implemented to meet the ROS messaging requirements.In terms of simulation on Matlab, the motion model of the robot was proposed based on kinematics, and the further research on motion control was also done. In additional, simulation on positioning based on dead reckoning and landmarks was verified. In the section of SLAM, analysis on EKF-SLAM and FastSLAM recursion algorithms was accomplished.In the respect of robotics theory research, inertia measurement data and odometry data were fused to make a better estimation of the robot state with Kalman filter. In this paper, the control model of the robot and depth measurement model were also deduced; the two models were the key components for the SLAM algorithms. For the SLAM implement, the type of grid maps was selected, and in-depth analysis was made for EKF-SLAM and FastSLAM. Finally the task of maps construction was accomplished using FastSLAM by experiment in laboratory.