Researchers at the University of Surrey have presented a PEnG system that can provide accurate positioning for autonomous vehicles even without access to a GPS signal. The results of their research are published in the journal IEEE Robotics and Automation Letters.
How PEnG works
The new technology is based on the analysis of visual data from satellite and street images. The algorithm works in two stages:
-
first limits the search to a specific street;
-
then refines the coordinates by estimating the position and orientation of the camera.
Thanks to this, even a standard monocular camera, which is installed in most cars, can become a powerful navigation tool.
Why is this important?
The technology reduces the error from over 700 meters to 22 meters. This is especially important in tunnels, densely populated cities or remote areas where GPS is unstable or unavailable.
PhD student Tavis Shore notes that most systems currently rely critically on GPS, while PEnG demonstrates the possibility of high accuracy without it. Project leader Simon Hadfield adds that this will make autonomous transport more reliable in real-world conditions.
Further steps
The team is already developing a prototype as part of a university grant program. According to Professor Adrian Hilton, the approach combines artificial intelligence and computer vision and has practical benefits for solving current problems in navigation.
