The Research On Adaptive Algorithm Of Mobile Robot Simultaneous Localization And Mapping

Mobile robot navigation is an important problem in the robot research field. Localization is the basic function that robot should possess, while at the same time, localization and mapping has a close relationship. In order to provide a mobile robot with truly autonomous capabilities, the robot localization and mapping (SLAM) must be seen as one problem to be solved. The success solution of SLAM is one of remarkable achievement in the robot field and has been widely used in outdoor, underwater and land environments. Faced with real environment with complexity and dynamic characteristics, in order to improve the autonomous capacity of mobile robot, the SLAM solution with adaptability, robustness and high efficiency is the research focus of the robot field.The paper focuses on the Research on adaptive algorithm of mobile robot simultaneous localization and mapping. To deal with the defects of traditional SLAM algorithms, the paper proposed the improved solution which makes the SLAM algorithms more adaptive and robust. The autonomous navigation capacity can be improved therefore. The main contributions of this thesis include the following aspects:First, the paper established the model of mobile robot navigation system. Main models that will be used in the research include: environment maps model; robot location and motion model; Sensor observation models; noise model, et al. The accurate modeling will establish appropriate platform basement for the SLAM research.Second, the paper researches some traditional SLAM methods. From the research, we know that extended kalman filter (EKF) is the traditional method of linearization of nonlinear system and it can solve the mobile robot simultaneous localization and mapping problem. The better aspect of particle filter (PF) is its high accuracy in state estimation, but while its problem is that the whole algorithm will generate large amounts of particles. So in order to achieve real-time navigation, the particle filter needs to be improved. Unscented kalman filter (UKF) describe the probability distribution of state vector by certainty sampling methods and it can avoid Particle degradation.Then, to deal with the problem of above algorithms such as bad robustness and adaptability, the paper proposed an adaptive SLAM algorithm based on strong tracking UKF. The advantage of STF is that its strong robustness of model mismatch; and also it has clear concept and low computational complexity. The improved SLAM algorithm combines the strengths of strong tracking filter (STF) and UKF. In the paper, each sampling point of UKF is updated by STF, and the filter gain was adjusted online. Therefore, adaptability and robustness was improved.At last, the paper summarized the whole research commitment. And also give the prospect of the mobile robot simultaneous localization and mapping.