Vehicle Classification and Speed Estimation using Magnetometers
Traffic Calming is a major aspect being studied in today's growing cities. The performance and efficiency of road layouts directly affect safety of drivers and the quality of life of the surrounding communities. Existing traffic monitoring solutions however, are either costly in terms of infrastructure and installation (inductive loops and pneumatic tubes) or privacy and placement sensitive (vision-based).
A low-cost, low-power alternative is to use MEMS-based magnetic field sensor that can easily be installed in pavement markers. Previous studies have shown that these magnetometers are sensitive enough to pick up characteristic magnetic profiles of vehicles moving past them. In this study, we will further examine these signals to determine vehicle speed and class using signal analysis and machine learning.
This study is conducted in collaboration with the City of Palo Alto to provide invaluable analytic data on their road networks. The developed algorithms will be deployed in low power, wireless embedded systems as part of a long term study to evaluate their efforts in improving the city's way of life.
Desired Outcomes for Semester
- Algorithms to determine vehicle speed and class using classical signal analysis or machine learning
- Evaluation of accuracy of using developed algorithms
- Deployment of algorithm(s) on embedded platform
- Evaluation of performance and power requirements to execute algorithm on embedded platforms
- Detailed technical report to include in conference paper
- Experience using Matlab
- Knowledge in the following areas:
- Signal Analysis ( using Wavelets )
- Machine Learning
- Experience in C/C++
- Version Control using Git
Send an email with your resume to:
- Cef Ramirez - email@example.com