In conditions of poor visibility, human drivers often struggle to safely navigate their vehicles, and AI drivers have even greater difficulties. This month, T&BB spoke to CEO and co-founder of Sensible 4, Harri Santamala, about the company and its product, DAWN – an autonomous driving technology specifically designed to deal with all weathers and all visibility conditions. 

Finland – One of the many questions which plague developers of autonomous driving technology initially appears to be very simple. How do you make a vehicle which can see the road ahead? The short answer is to give it artificial eyes: for example, cameras facing all around the vehicle; or ‘lidar’, which uses amplified light (i.e., lasers) to accurately judge distances between the vehicle and other objects. Many prototype vehicles use a combination of both. However, a vehicle cannot “see” in any meaningful sense unless it is also able to “perceive” what it can see. The act of seeing is never simply a matter of looking at something; the observer is always making distinctions between objects and interpreting them. In the driver’s situation, accurate perception is essential. Is that another vehicle or a pedestrian? What is its direction and its speed? Do I need to manoeuvre to avoid a collision? It is the job of the software embedded in autonomous driving systems to come up with answers to these questions in real time, and it can give correct ones only if it can perceive and interpret the objects surrounding the vehicle.