Previous Abstract Return to Session C4 Next Abstract

Session C4: Urban and Indoor GNSS

Set-Based Ambiguity Reduction in Shadow Matching with Iterative GNSS Pseudoranges
Daniel Neamati, Sriramya Bhamidipati, and Grace Gao, Stanford University
Date/Time: Thursday, Sep. 22, 2:35 p.m.

Best Presentation

3D map-aided GNSS localization provides state-of-the-art urban positioning by exploiting 3D building maps to account for reduced satellite visibility. Shadow matching is at the core of 3D map-aided GNSS but yields a multimodal posterior distribution due to symmetries in building geometry that degrades localization. Past works used the GNSS pseudorange with grid-based discretization to address this multi-modal ambiguity. However, grid-based formulations are not suitable for all users since they require fixed resolutions and discretized state representations. Instead, we propose a new set-based solution to incorporate GNSS pseudoranges in an iterative filter to support our prior work on set-based shadow matching. We take advantage of the four-dimensional conic geometry of the pseudorange equation to enable set-based operations. We then design a backward-forward filtering architecture that iteratively identifies the most probable shadow matching mode. We validate our approach on real-world smartphone data collected in an urban area of San Francisco. Our method successfully identifies the correct shadow matching mode with high confidence across all timesteps, compared to shadow matching alone or a single-step fusion of shadow matching and set-based pseudoranges. We demonstrate that our method maintains its identification rate and computationally efficiency, even for a higher number of ambiguous modes. With vectorizable set-based operations, the full filter runs in 5.4 ms per timestep for six shadow matching modes. Hence, our method is accurate and suitable for real-world operations.

Previous Abstract Return to Session C4 Next Abstract