SLAM method based on augmented EKF for mobile robot
XIAO Xiong1,2, LI Dan1,2, CHEN Xiduan1,2, LI Gang1,2
(1.College of Information Engineering, Zhejiang University of Technology, Hangzhou 310023, China;
2.Zhejiang Key Laboratory for Signal Processing, Hangzhou 310023, China)
Abstract: Aiming at the problem that the accumulation of systematic odometry error in the process of simultaneous localization and mapping (SLAM), by adopting the relationship of the odometry error model mapped to velocity error model of each wheel and combining augmented extended Kalman filter(AEKF) algorithm structure and considering reality robot model, one SLAM method efficiently improving the precision of localization was proposed. The systematic velocity calibration parameters were appended to the state vector of EKFSLAM algorithm becoming an augmented state, and then these parameters and the SLAM initial vector were predicted and updated. Through revising the robot′s velocity and orientation online, the orientation error and odometry error were decreased. The root mean squared error(RMSE) and normalized estimation error squared(NEES) were tested. The results indicate that, comparing with conventional EKFSLAM, the proposed method has better estimation performance, keeps the algorithm consistency and generates more accurate robot localization and feature map.
Key words: augment extended Kalman filter; simultaneous localization and mapping(SLAM); odometry error; root mean squared error; normalized estimation error square