Twofeet : an infrastructure-free localization system for large indoor spaces and its applications using smartphones
This project implements TwoFeet, an infrastructure-free localization system designed for turn-by-turn navigation within large indoor spaces. While current approaches exist, many of these contending systems necessitate a complex infrastructure and do not take into consideration global scalability, en...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
2013
|
| Subjects: | |
| Online Access: | http://hdl.handle.net/10356/51959 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | This project implements TwoFeet, an infrastructure-free localization system designed for turn-by-turn navigation within large indoor spaces. While current approaches exist, many of these contending systems necessitate a complex infrastructure and do not take into consideration global scalability, energy efficiency, cost and ease of deployment.
Utilizing solely the on-board accelerometer and magnetometer sensors available in modern smartphones, TwoFeet accurately localizes a user along a recommended shortest route. However, implementing this technique in human-scale environments is non-trivial as noise inherent to mobile sensors and complicated human kinetics conjure real world research problems. In order to compensate for inaccuracies arising from these challenges, an intelligent algorithm was designed to match detected user activity onto path signatures. This project also aimed to plug a gap left by similar systems that do not deal with user deviating from the displayed route. As part of the localization approach, a simple and inexpensive procedure was developed to map indoor environments without requiring expensive calibration efforts. Lastly, to provide a value-added experience for mobile users, this project also explored the integration of location-relevant features such as in-navigation advertising, crowd-sourced queue time information for F&B outlets and Where did I Park?.
The resulting implementation was tested in a sample shopping mall, of which good localization precision and robustness was achieved. TwoFeet exhibited a 40% improvement in average accuracy compared with its closest related approach as well as successfully detected, bridged and corrected all cases of user divergence. |
|---|