Events

PhD Defense: "Localization and Mapping using GNSS SNR measurements"

Andrew Irish

April 26th (Friday), 10:00am
Harold Frank Hall (HFH), Room 4164


Recent advances in technology, in particular the miniaturization and commoditization of electronic wireless receivers, have enabled new applications and techniques for wireless communication, mapping, and localization. The example we provide is focused on GNSS (Global Navigation Satellite System, e.g. GPS) receivers, which widely deployed in mobile consumer electronics for positioning, can be also be leveraged as environment sensors. The problem of estimating a 3D map, or a world model, via crowd-sourced wireless signal strength measurements from global positioning satellites, is explored in detail. The inverse problem — localizing the receiver against a known or partially known 3D map — is then discussed, as is the fully integrated simultaneous localization and mapping (SLAM) problem where both the trajectory of the GNSS receiver and 3D map are jointly estimated. For all of of these cases, novel probabilistic modeling approaches of the map and receiver location, as well as the channel model, are introduced. The proposed solutions are validated using real-world data from consumer-grade mobile devices (smart phones) on both small and larger (multi-city block scale) experiments. For positioning, the backend positioning server and associated smartphone app were prototyped in software, with improved localization using a particle filter being demonstrated in real-time.

About Andrew Irish:

photo of andrew irishAndrew Irish is a PhD candidate in the Wireless Communication and Sensornets Laboratory (WCSL) under Professor Upamanyu Madhow. While conducting the initial GNSS SNR positioning research, Andrew and his colleagues Danny Iland and Jason Isaacs won the Grand Prize in the 2014 UCSB New Venture Competition showcasing ShadowMaps, which applied cloud computing on GNSS SNR measurements from smartphones to improve GNSS (GPS) positioning accuracy in urban canyons. ShadowMaps was acquired by Uber in 2016, and Andrew, Danny and Prof. Madhow joined Uber, where they fully deployed ShadowMap’s location improvement technology in order to improve the pickup experience for riders. Today, Andrew is still employed at Uber with his colleague Danny, who are Senior Software Engineers on the Location Intelligence team.

Hosted by: Committee: Upamanyu Madhow (chair), Joao Hespanha, Yasamin Mostafi, Heather Zheng