IPAB Workshop - 13/11/25

Title: Why SLAM? Optimal Robot Localisation without Mapping in Unknown Environments

Abstract: Simultaneous Localisation and Mapping (SLAM) is a fundamental problem in autonomous robot navigation in unknown environments. It requires the robot to use the information obtained from its onboard sensors to construct a map of the unknown environment and, at the same time, localise itself within the map, particularly in the absence of an external location reference such as a global positioning system (GPS). In our recent work, we proved that in 2D and 3D feature-based SLAM, when the covariance matrices of feature observation errors are isotropic, solving robot poses (“localisation”) is completely independent of features (“mapping”). This independence property contradicts the well-accepted belief that “localisation and mapping need to be performed SIMULTANEOUSLY in unknown environments”. This talk will first introduce the proof of the independence property in feature-based SLAM, and then show some equivalent optimal localisation-only algorithms for different SLAM problems based on this independence property.