Multipath-Based SLAM With DOA Ambiguity

Multipath-based simultaneous localization and mapping (SLAM) relies on high-resolution direction of arrival (DOA) estimates to achieve excellent performance. Expanding antenna element spacing is a common strategy to improve DOA resolution. However, excessively large spacing can cause signals from different directions to become indistinguishable, leading to DOA ambiguity. Traditional methods typically address DOA ambiguity through antenna array relocation or the use of multiple arrays. In multipath-based SLAM, however, the movement of the mobile agent naturally facilitates the movement of the receiver's array, allowing the use of high-resolution yet ambiguous DOA estimates. In this paper, we propose a splitting state transition model to handle DOA ambiguity and a novel amplitude measurement model to handle the high nonlinearity it introduces. Integrating these models within an existing Bayesian framework, we develop a message passing algorithm. Simulations demonstrate that the proposed method improves positioning and mapping performance, while exhibiting enhanced robustness against divergence.