SLAM mais Preciso utilizando Filtro Controlado Aprimorado por Alinhamento Ativo

Abstract

This work presents a novel method, called Controlled Filter with Active Alignment (CFAA), to address the Simultaneous Localization and Mapping (SLAM) problem. SLAM aims to map an unknown environment while estimating the trajectory of a mobile agent moving within that environment. CFAA integrates the two fundamental pillars of SLAM, scan alignment and loop closure, into a unified process. The method draws inspiration from human perception of self-localization and employs a mental map for guidance. CFAA employs a Gaussian distribution to estimate potential poses and carries out the alignment process in cycles, where each cycle is influenced by the outcomes of the preceding one. The Active Alignment mechanism is employed in each cycle to determine the quality of each potential pose, enabling more accurate simultaneous localization and mapping. Benchmark tests were conducted on five public datasets, verifying the effectiveness and efficiency of the proposed method.