Colored-State-Noise Simultaneous Localization and Map Building for Wheel Robots

Abstract

Since the control state of wheel robots from encoder is colored noise in the mechanical manufacture,the traditional simultaneous localization and map building algorithms are on longer applicable.In the paper,an algorithm of colored-state-noise SLAM is proposed.Nonlinear process model is linearized and colored-state-noise model is converted into Gauss white noise one by augmenting dimension of state.The integral algorithm procedure follows the recursive order of prediction,observation,data association,update,mapping to have simultaneous localization and map building.MonteCarlo simulation results show that the proposed algorithm has higher estimate precision than those of EKF-SLAM algorithm and Fast-SLAM algorithm in colored process noise covariance.