Research On The Problem Of Localization And Mapping For A Mobile Robot

As the representation of the highest level of automation, robotic had collected a large amount of newest results of science and technology. Mobile robot is an important branch of robotics, and many researchers had focused on it for long time. Mobile robot will also come as the destination of robotics. Navigation of the robot is the premise of mobility of it, and in which mapping and localization is the basic skill of the robot. Goal of Mapping and Localization of robot is to ensure the autonomy exploring of the surroundings and awareness of localization of itself which is important to navigation.Inner sensors such as encoder and electronic compass are applied to sense the motion state of the robot and laser range finder, ultrasonic sensor and node of the wireless sensor network (WSN) are applied to sense the surroundings as external sensors. Mapping and Localization of the robot refers to the process that the robot localization itself and sensing the surroundings combining the information of itself and surrounding reading from the sensors. This paper is placed around the problem of localization via WSN, mapping via ultrasonic sensor and Simultaneous Localization and Mapping (SLAM) based of ultrasonic sensors. Works of this thesis come as the following:Firstly, an algorithm based on the prediction of Received Signal Strength (RSS) of the WSN via the motion model of the robot is presented, which is focused on the accurate localization of robot in the complex indoor environment with Non-line of sight (NLOS). RSS is predicted and modified via the motion model of the robot, and NLOS is judged via the results of predication. Extended Kalman Filter (EKF) is applied for optimizing of the result of localization. A novel strategy of localization based on WSN is presented based on the Gaussian model. Distances between the mobile node and reference nodes are estimated via RSS and contact points of each two circles are calculated. Gaussian model is applied to gain the expectation of coordinate of the mobile node. Also EKF is applied for optimizing and NLOS is judged via motion model and result of localization. Result of the presented algorithm show us that it improved the accuracy of localization.Secondly, an algorithm for line extraction of line features via ultrasonic sensor data is presented based on the middle line model of the ultrasonic sensor. Reason of the uncertainty of direction of the ultrasonic sensor is analyzed and a model of ranging via ultrasonic sensor is presented in order to reject the wrong returns from the sensor. An improved Iterative End-Point Fit (IEPF) is presented to extract line features, and EKF is applied to estimate parameters and fusion of them. An algorithm for mapping based on the extraction of Region of Constant Depth (RCD) is presented based on the Gaussian model of the sensor. Gaussian model is applied to the two measurement correlation model of RCD, and Least Square Method (LSM) is applied to estimate the parameters of the features. Methods for matching and fusion of the features are also presented.Thirdly, a deep insight of SLAM for mobile robot is presented. An algorithm based on imposing restriction on observability and additional estimation of pose is presented, in which observability of EKF-SLAM was analysed both in version of linearized and non-linerized. An algorithm for ensurance of accordance of the observability of linearized EKF-SLAM to non-linearized EKF-SLAM system was presented which can contribute to the consistency of EKF-SLAM. EKF-SLAM with additional estimation of pose (AE-EKF-SLAM) is presented in which the observations are applied for pose estimation before EKF, which improves the accuracy of the linearizing of the non-linear system of SLAM. Simulation results show that the presented AE-EKF-SLAM improves the accuracy of SLAM of robot.At last, details of the design of mobile robot platform are presented on which experiments show in this paper is done. Power requirement of platform is presented before hardware design, and also the configuration of internal and external sensors are presented. Motion control of the platform and stroages and collection of data are introduced in design of software, and also the transmission of data via wireless method is designed. It is certificated that the platform works well in the experiments in this paper, which can be a foundmental tools of our researchs.