Localization And Mapping Indoor For Mobile Robot Based On Three-dimensional Scanner

Three-dimensional perception and localization in unknown environment is the foundation of many robotics applications and of great significance to improve the robot's autonomy. Simultaneous localization and mapping based on laser with high precision, anti-interference and robustness become a hotspot in the field of robotics. We design a system based on self-developed 3D laser scanner, MEMS inertial measurements and the odometer, which can accurately estimate the six degrees of freedom trajectory of a sensor platform, as well as a 3D point cloud map of that environment. This paper is organized as follows.I) Analyze the error model of MEMS inertial sensors and develop attitude measurement model based on Kalman filter, then design the localization subsystem combined with odometer.II)The NDT and ICP registration algorithm are compared with each other and the NDT registration algorithm is more quickly and accurately indoor. Registration of point cloud is integrated with odometer localization subsystem, which accelerate the registration and reduce the accumulated error of odometer subsystem.III)In order to reduce the registration accumulated error, it uses NDT loop detection algorithm to detect loop closure and form a pose graph, then a pose graph optimization method is applied to the pose graph.IV)To verify the method which have been described above, it implements the inertial sensor with odometer localization subsystem based on STM32F407, and proposes a calibration method for the misalignment between inertial sensors and three-dimensional scanner. The Qt PC software is completed for data visualization and local-saved.Finally, this system is proved to be stable, simple and high precision, which fit for the diverse indoor environment and reduces the cost of localization system based on laser.