Grant: Alliance Grants

With the improvement in hardware and the ability to deal with a large amount of data, cooperative control of multiple vehicles became feasible, either using traditional model-based techniques or modern data-driven approaches based on machine learning and artificial intelligence. However, interactions among agents are still hard to model and may lead to instability in the execution of mission, especially in dynamic environments or when vehicles cannot have access to all the states of the system due to limitations in sensing and/or communications.This project aims to study the interactions among diverse vehicles and develop solutions for robust task assignment that considers the motions and dynamic restrictions of vehicles. The solutions will use a mixture of traditional modelling and data-driven techniques based on machine learning approaches, mostly deep reinforcement learning. The dynamical modelling of vehicles will be done using topological approaches in conjunction with Game Theory concepts. Furthermore, optimization of task assignment will be based on mechanism design and reinforcement learning. Scenarios based on diverse settings, mostly in the infrastructure (power generation and transmission) and defence sectors will be developed.The results will be transferred to the partners for further development. It is expected that the solutions will enhance the market position of the partners and improve the competitiveness of Canadian industry.