In this research proposal, a cooperative decentralized adaptive model predictive control (MPC) methodology will be developed. The proposed methodology will incorporate dynamic models produced using computational intelligence (CI) techniques and, more specifically, by radial basis function (RBF) neural networks.
MPC methods are usually implemented through the logic of centralized control, according to which the controller has access to the measurements of all the nodes that comprise a system at any given time. Unfortunately, centralized control methods present significant disadvantages, such as the increased computational costs and the risk of complete system failure in the event of controller malfunction, or loss of communication with the nodes. In the current project, emphasis will be given in decentralized control, where each node 
solves an individual optimization problem according to its own personal goal, while also considering the state of neighboring nodes. Thus, the total computational cost is broken down to a number of smaller optimization problems, while the success of the control system does not depend on the communication with a central node.
An important novelty in the field of decentralized control will be the integration of these models in adaptive MPC controllers, so that the objective function and the restrictions of the optimization problem that each node separately solves will include a prediction of the neighboring nodes behavior, thus creating a cooperative control framework. This way the “selfish” behavior of the nodes will be avoided and cooperation will be achieved producing an overall better result compared to the case of simple distributed control.
The proposed methodology will be applied to problems of cooperative decentralized control of aerial vehicle swarms, as well as fleets of surface vehicles. These applications are expected to provide significant benefits in terms of traffic control at points of interest, such as airports, ports, and canals, but also in assignments of surveillance, exploration and transport of goods and passengers. Some of these benefits target the reduction of fuel consumption and travel time through the calculation of the optimal trajectory, while also increasing navigation safety.
Project title: “Cooperative distributed adaptive model predictive control methods using computational intelligence”
Cooperating entities: The project is implemented by:
Laboratory of Telecommunications, Signal Processing and Intelligent Systems, Department of Electrical and Electronic Engineering, School of Engineering, University of West Attica.
Source:
This research is co-financed by Greece and the European Union (European Social Fund- ESF) through the Operational Program “Human Resources Development, Education and Lifelong Learning 2014-2020” in the context of the project MIS 5050291.
Funding amount:
Total budget: 41.041,00€
Duration: 01/05/2020 – 31/07/2021
Principal Investigator: Dr. Alex Alexandridis, Professor, Department of Electrical and Electronic Engineering, University of West Attica, Office: A115, Campus 2, Office Telephone Number: 2105381571
