The Cambridge-based start-up PROWLER.io has raised £10 million in a series A funding round led by Cambridge Innovation Capital (CIC) to tackle what it calls the challenge for developers of autonomous vehicles to ensure they operate effectively in the real world.
PROWLER.io says its novel decision-making platform combines for the first time three different strands of artificial intelligence: probabilistic modelling, reinforcement learning and game theory.
Vishal Chatrath, CEO of PROWLER.io says that initially people thought he was “bonkers”, but now he has brought together a team of world leading experts that share his vision and taking the example of a self-driving car, he explains there are three main elements to the problem, “The car needs to be able to “understand” itself and the roads, it needs to respond appropriately to changes in the environment and it needs to interact safely with other vehicles and humans. This creates a multidimensional problem that is constantly changing,” he says.
“Artificial Intelligence can be split into three camps, those that attempt to classify everything (it’s a bike, not a car), those that create decision trees (if the road is icy, brake carefully) and those that manage interactive behaviour (stop and let the pedestrian cross),” he adds. “We argue that you need all three strands of AI – probabilistic modelling, reinforcement learning and game theory – to work together to understand and control all the millions of micro-decisions that occur in complex systems.”
Andrew Williamson, Investment Director at CIC says: “PROWLER.io is a developing a brand new strand of AI that can be applied to a wide range of intractable problems. It is rare to see such as leap forward in technology, but it is based on a rich legacy of academic work that has been nurtured in Cambridge for over 10 years. Chairman Professor Carl Rasmussen literally wrote the book about probabilistic modelling!”