Control of connected vehicle platoons can ensure the swift movement of traffic through a city by sharing vehicles’ states and desired actuation. This networked control design can alleviate traffic jams, reduce vehicle emissions, and reduce fuel usage through improved aerodynamics. Model Predictive Control algorithms are a natural solution to address constraints arising from both communications and system dynamics. A key challenge is to design distributed control algorithms that are robust to disturbances in the environment and to stochastic information from the communication network.
This PhD project aims to develop new knowledge in model predictive control algorithms that incorporate stochastic communication network effects, e.g., reliability, dropouts, information delays, for the distributed network of agents. It forms part of a larger research endeavour jointly with Prof. Falko Dressler from TU Berlin and Prof. Rolf Findeisen from TU Darmstadt and will aid to the fundamental engineering science needed to bring automated vehicle systems into practice.
Also involving Justin Kennedy
Two related publications:
- Stabilizing stochastic predictive control under Bernoulli dropouts, IEEE Transactions on Automatic Control, 2018
- Predictive control for networked systems affected by correlated packet loss, International Journal of Robust and Nonlinear Control, 2019