Data-driven control of autonomous vehicle and fleets

Next generation 5G networks will bring about a new digital revolution, enabling data rich applications such as the internet of things, and large-scale wireless sensor networks. Massive data produced by these infrastructures will require in turn capabilities of exploiting efficiently the information contained in the data. There are countless of potential applications, ranging from localization and control of autonomous vehicles and fleets. The proposed research activity will indeed regard this latter theme (control of autonomous vehicle and fleets). In this context, a data-driven approach will be adopted, where the required models and control algorithms will be synthesized from the available data, whithout requiring to use detailed “first-principle” models, that in this kind of applications may be quite complex and difficult to derive. Another advantage of this approach is that the resulting control algorithms can be more efficient from a computational point of view with restpect to “first-principle-based” algorithms, thus allowing an on-line effective implemetation. The main lines of the research activity will include distributed estimation, motion prediction, trajectory planning and decision making. The estimation and control problems arising in these research lines will be tackeled by means of recently developed data-driven  methods, such as direct virtual sensor (DVS) estimation and data-driven model predictive control (MPC). 

 ERC Sector:

  • PE7_1 Control engineering
  • PE6_11  Machine learning, statistical data processing and applications using signal processing

Keywords:

  • Control systems
  • Autonomous vehicles
  • Data modeling
  • Optimization
  • Machine Learning
  • Networked systems
  • Distributed estimation
  • Direct Virtual Sensors
  • Model Predictive Control

Gruppi di ricerca