Dr Ben Metcalfe, a lecturer in the Department of Electronics and Electrical Engineering at the University of Bath, UK, predicts how automotive testing and development will have changed in 20 years’ time
One of the greatest opportunities, and the most significant challenges for the future of automotive development will be the intelligent interaction between vehicles. Modern sensor technologies such as lidar and optical SLAM have made it possible for a vehicle to build a real-time three-dimensional map of its environment. This can then be used in scenarios such as automated collision avoidance or adaptive cruise control. This paradigm can be extended even further: if vehicles are given the ability to communicate directly with one another, then they might share information such as current location, speed, road conditions and destination. Now imagine this level of communication coupled with autonomy – wave goodbye to traffic lights and stop signs, and traffic arbitration could be performed in real time by the cars themselves.
This kind of technology is already introducing new challenges in the testing and development of modern vehicles. Optical camera systems for example require large data sets for training and testing purposes, these need to encompass a broad range of scenarios such as different lighting, weather patterns, vehicle types, speeds and road conditions. There is a greater need for simulated environments that can replicate the real-world conditions as accurately as possible.
When vehicles are able to talk to each other then these challenges increase further, as collective – or swarm – decisions need to be made to optimize traffic flow. The simulated environments required for testing and development would need to also support multivehicle scenarios that account for a mix of autonomous and traditional vehicles. There are also security implications, as false data transmitted between vehicles could be used to intentionally cause crashes or traffic chaos.
The potential for economic and environmental benefit by developing collective autonomy in future vehicles is obvious; the challenge is how to develop and validate such a complex system. No doubt the rise of machine learning and artificial intelligence will play a significant role, but there is plenty more work to do yet.