Anytime, Anywhere Low Cost Localisation

Being able to precisely answer the question of "Where am I?" is critical for autonomous vehicle navigation - a function known as "localisation". There are a number of ways that a vehicle can localise: while GPS is an example of a localisation system, it is insufficiently accurate for autonomous driving systems, as well intermittently available and susceptible to jamming and interference. Lidar, a laser-based scanning technique, is commonly used to provide estimates of localisation to driverless cars, but lidar sensors are too costly for mass market vehicles. Cameras are significantly cheaper than lasers, but image-based localisation is challenging because of changes in lighting, weather, and scene structure. Taking into account the pros and cons of each of the above methods, this joint Ford-Oxbotica project utilises a suite of innovative techniques to perform camera-only localisation in spite of these environmental changes. The project will trial the software using low cost hardware to demonstrate the performance of affordable technology for mass market adoption.