A Bayesian occupancy grid filter for robust pedestrian dead reckoning

This thesis considers the problem of indoor localization using inertial sensors (IMUs) that are embedded in almost all smartphones and mobile devices. Significant research has focused on developing pedestrian dead reckoning (PDR) algorithms that utilize the IMU measurements to track human movement. While Particle Filters (PF) have offered the best-known accuracy so far, they are also known to suffer from low robustness. This is not surprising given how IMU data from the real world is highly noisy, mainly due to the arm and limb gestures of the user. Since real-world deployments often favor robustness over accuracy, I propose a Bayesian Occupancy Grid Filter (BOF) that can absorb far greater IMU error compared to PFs. The robustness gains are shown through extensive simulations and the implementation of a fully functional real-time system that performs in accordance with our expectations. BOFs are also simple to implement and can be an important step toward the wide-scale deployment of IMU-based indoor positioning systems.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Electrical & Computer Engr

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2023

Author dc:creator

• Bhandary Karnoor, Sahil

Contributors dc:contributor

• Roy Choudhury, Romit

• Bayesian Filtering
• Indoor Navigation
• Pedestrian Dead Reckoning
• Inertial Measurement Units

Statement dc:rights

• Copyright 2023 Sahil Bhandary Karnoor

Language dc:language

en, eng

Handle dc:identifier

https://hdl.handle.net/2142/122157

OAI record oai:identifier

https://www.ideals.illinois.edu/oai-pmh?verb=GetRecord&metadataPrefix=etdms&identifier=oai%3Awww.ideals.illinois.edu%3A2142%2F122157