Planning under uncertainty for coordinating infrastructural maintenance

Abstract

Scheduling of infrastructural maintenance poses a complex multi-agent problem. Commonly a central authority is responsible for the quality and throughput of the infrastructure, while the actual maintenance is performed by multiple self-interested contractors. Not only does the central authority have to (economically) incentivise agents to consider quality and throughput, it is also burdened with the coordination of agents' activities on the network with contingent activity durations. We introduce a coordination method that combines planning under uncertainty and dynamic mechanism design to coordinate agents on a network level. We apply this method on maintenance planning scenarios obtained through accurate modelling of the problem domain. To the best of our knowledge, this is the first application of dynamic mechanism design on a real-world problem. Finally, we validate the feasibility of our method through experimental evaluation and identify current open challenges for both the planning and scheduling as well as the mechanism design communities.