Self-learning software to make train maintenance more efficient This project will research into how AI techniques, which can quickly derive plans to tackle complex problems, could be used to better plan train shunting movements. It will also examine how these techniques could take account of human planners’ existing preferences and how these plans, if necessary, could be efficiently adapted. It is expected that the use of AI will raise the capacity of shunting yards, allowing more trains to be parked and maintained there.