Research Of Mobile Robot Localition And Map Building Based On Improved CKF

Robotics as a discipline crossed by automatic control technology, sensor technology, microprocessor technology, artificial intelligence and other fields, relates to human life and work, represents a country’s comprehensive strength and the level of scientific research, caused by the national research institutions and high-tech enterprises researching and exploring, the research achievements of robotics has far-reaching significance and is inseparable of civilization.The application environments of some robots are unknown, so how to locate and build map in unknown environment is the key factor of those kinds of robot, the SLAM problem of mobile robots.In this paper,improves the existing algorithm CKF-SLAM and optimizes volume sampling rules and reduces the influence of initial error and linear error of state estimation, in order to generate the posterior probability distribution is more close to the true value. Establish the corresponding experimental platform of mobile robot and validate the corresponding algorithm. This paper mainly includes:First, Establish the environmental coordinate system of a mobile robot, integrate environment coordinate system and mobile robot coordinate; elaborate the methods and steps of the feature map in complex environment; aiming at the problem of data processing and sensor data acquisition,describe the motion model, the observation model and the noise model of the mobile robot in detail.Then, introduce the Kalman filter framework, describe the principle of the classical Extended Kalman filter (EKF) and Unscented Kalman filter (UKF), and discusse the advantages and defects of the two algorithms,simulate EKF-SLAM algorithm and UKF-SLAM algorithm in the experiment, analyze the results.Continuedly, study the mathematical principle of cubature Kalman filter, for the square root cubature Kalman filter algorithm in mobile robot map building and localization problem exists with the increased map feature point, volume points deviate from the ideal track and state estimation has large error.For the factor, an improved square root volume Kalman filtering algorithm is proposed. The algorithm introduces an iterative measurement update method, in the update phase using the estimated value and square root factor to determine the sampling volume and makes the sampling points in a highly nonlinear environment keep little distortion, improving the accuracy. Simulation results show that the proposed algorithm can improve the position and pose accuracy of the robot, compared with the square root volume Kalman filter algorithm.A mobile robot experimental platform is built according to the experimental need. The experimental platform of mobile robot pose to the Traveller Robot 2 based and consists of the sensor module, network communication module, input module and so on.Among them, sensor module mainly includes:laser radar unit and Photoelectric coding unit; network communication module contains a wireless network card and an onboard computer; upper machine and lower machine synergy form the input module of computer and PC that is onboard computer, the main task is self localization, scene construction, path planning; lower machine is control board, responsible for docking sensor module and realize control function. Some experiments were carried out on the experimental platform to verify the actual performance of the SLAM algorithm.