Robust network localization for teams of robots in non-line-of-sight environments

Motivated by collaborative localization for multi-robot teams, we consider the problem of large-scale network localization. We present Sparse Matrix Inference and Linear Embedding (SMILE) a novel approach to network localization which draws on both the well-known Locally Linear Embedding (LLE) algorithm and recent advances in sparse plus low-rank matrix decomposition. We demonstrate that our approach is robust to noisy pairwise distance measurements and the severe effect of non-line-of-sight signal propagation. Our experiments include simulated medium and large-scale networks as well as data from an 11 node cell phone Bluetooth LTE network and an 18 node network of mobile robots and static radios in a GPS-denied limestone mine. Our findings indicate that this method outperforms a state-of-the-art graph convolution network-based approach in terms of both localization accuracy and compute time.