Final Year Project.
Shared mobility has experienced significant growth in recent years and has the potential to reduce the climate impact of transportation through a more efficient use of resources in terms of vehicle utilisation, road space and energy.
On-demand services have been offered for a wide range of vehicles including cars, vans, bicycles, electric bicycles and electric scooters.
For these services to offer a compelling service, competitive with individual vehicle ownership, they need to offer reliable availability of their vehicles to customers.
Customer preferences however may result in asymmetric demand for journey origin and destination points which can lead to a deficit of vehicles in one location and a surplus at another.
Service providers can counter this by redistributing their vehicles but of course these redistributing efforts come with costs.
This project builds on a prior project to explore the application of Reinforcement Learning to vehicle rebalancing decisions. An openai gym environment will be used to test the effectiveness of a range of Deep Reinforcement Learning techniques to optimising redistributing efforts across a range of conditions.