A multi-innovation with forgetting factor based EKF-SLAM method for mobile robots

Abstract

Purpose The purpose of this paper is to explore a multi-innovation with forgetting factor-based EKF-SLAM (FMI-EKF-SLAM) algorithm to solve the error increasing problem, caused by the Extended Kalman filtering (EKF) violating the local linear assumption in simultaneous localization and mapping (SLAM) for mobile robots because of strong nonlinearity. Design/methodology/approach A multi-innovation with forgetting factor-based EKF-SLAM (FMI-EKF-SLAM) algorithm is investigated. At each filtering step, the FMI-EKF-SLAM algorithm expands the single innovation at current step to an extended multi-innovation containing current and previous steps and introduces the forgetting factor to reduce the effect of old innovations. Findings The simulation results show that the explored FMI-EKF-SLAM method reduces the state estimation errors, obtains the ideal filtering effect and achieves higher accuracy in positioning and mapping. Originality/value The method proposed in this paper improves the positioning accuracy of SLAM and improves the EKF, so that the EKF has higher accuracy and wider application range.