Optimizing the efficiency of the sub-map technique for large-scale simultaneous localization and mapping

A technique for optimising the efficiency of sub-map method for large-scale simultaneous localisation and mapping (SLAM) is proposed. It optimises the benefits of the sub-map technique to improve the accuracy and consistency of an Extended Kalman Filter (EKF) based SLAM. Error models were developed and engaged to investigate some of the outstanding issues in employing sub-map technique in SLAM. Such issues include: the size (distance) of an optimal sub-map, the acceptable error effect caused by the process noise covariance on the predictions and estimations made within a sub-map, when to terminate an existing sub-map and start a new one and the magnitude of the process noise covariance that could produce such an effect. Numerical results obtained from the study and an error correcting process was engaged to optimise the accuracy and convergence of the Invariant Information Local Sub-map Filter previously proposed. Applying this technique to the EKF-based SLAM algorithm: (a) reduces the computational burden of maintaining the global map estimates and (b) simplifies transformation complexities and data association ambiguities usually experienced in fusing sub-maps together. A Monte-Carlo analysis of the system is presented as a means of demonstrating the consistency and efficacy of the proposed technique.