Search papers, labs, and topics across Lattice.
This paper introduces a novel message passing algorithm that integrates Gaussian belief propagation and mean-field approximation to enhance collaborative localization among multiple robots. By employing a normalizing flow-based gradient estimator, the method effectively addresses non-conjugate terms from nonlinear measurement models, allowing for learnable sampling and improved estimation performance. Experimental results on autonomous surface vehicles show significant gains in accuracy, robustness, and adaptability in localization tasks involving various sensor inputs.
Normalizing flow-enhanced message passing boosts multirobot localization accuracy and adaptability, outperforming traditional methods in real-world scenarios.
Accurate, robust, and adaptive localization is essential for various robotic operations. This paper proposes a new message passing (MP) algorithm for realizing collaborative localization in a distributed manner. The algorithm unifies Gaussian belief propagation (GBP) and mean-field (MF) approximation, where GBP preserves dependencies among robot states, and MF enables estimation of noise statistics. To effectively handle non-conjugate terms from nonlinear measurement models, the algorithm adopts a parametric formulation in which these terms are treated by gradient estimators. Beyond linearization and sampling, we further design a normalizing flow (NF)-based gradient estimator, enabling learnable sampling. End-to-end training tunes NF parameters according to the behavior of MP, improving the overall estimation performance. To support estimation of practical robotic states that involve rotations, the method is then extended to Lie group state spaces. Finally, the method is applied to multirobot localization task fusing odometry, global navigation satellite system (GNSS) measurements, and inter-robot ultra wideband (UWB) ranging. Simulations and experiments on autonomous surface vehicles (ASVs) demonstrate its improved accuracy, robustness, and adaptability.
