Evolving intelligent vehicle control using multi-objective NEAT

The research in this paper is inspired by a vision of intelligent vehicles that autonomously move along motorways: they join and leave trains of vehicles (platoons), overtake other vehicles, etc. We propose a multi-objective algorithm based on NEAT and SPEA2 that evolves controllers for such intelligent vehicles. The algorithm yields a set of solutions that embody their own prioritisation of various user requirements such as speed, comfort or fuel economy. This contrasts with most current research into such controllers, where the user preferences are summarised in a single number that the controller development process should optimise. Having multiple prioritisations of preferences would, however, allow the user to select desired vehicle behaviour in real time, for instance fast driving if she´s in a hurry or economical driving in more relaxed circumstances. Preliminary results of our experiments show that evolved controllers substantially outperform the human behavioural model. We show that it is possible to evolve a set of vehicle controllers that correspond to different prioritisations of user preferences, giving the driver, on the road, the power to decide which preferences to emphasise.