The goal of this project is to estimate the location of vehicles inside a tunnel in the absence of GPS reception.
Abstract
The goal of this project is to estimate the location of vehicles inside a tunnel in the absence of GPS reception. When the vehicle exits the tunnel its estimated location will be used for continuous navigation without waiting to reestablish GPS reception, which may take a substantial amount of time.
We propose a stand-alone solution running on a smartphone device that applies image processing algorithms on data received from the smartphone’s built-in video camera and processed quickly before new data arrives. The solution we offer, which is explained in detail in this book, is based on the fact that each tunnel has a known structure with lights and road marks from which we can derive the speed of the vehicle and its location. We estimate vehicle speed by identifying lamps along the tunnel, under the assumption that they are separated by known fixed distances.
We also find absolute position by identifying signs along the tunnel (digits indicating stopping points, only relevant to Carmel tunnels – which we focused on). Due to the system’s run-time limitation, we only used one third of the frames provided by the camera for image processing.
All experiments were conducted in the Carmel Tunnels in Haifa, and the results were satisfactory: At the end of the tunnel we received an estimation of the location with a negligible error of about 1-2 % and the speed estimation matched the actual speed sufficiently (we had no tools to measure exact vehicle speed).
Useful Acronyms
PTAM – Parallel Tracking and Mapping
SLAM – Simultaneous localization and mapping
MMSE – Minimum mean square error
SVM – Support vector machine
FPS – Frames per seconds
PTAM results
Identifying location using light bulbs
Final Results
