PhD position on Verifiable estimation and control for autonomous systems

Updated: about 1 month ago
Deadline: 25 Apr 2021

Emerging applications in robotics, aerospace and autonomous driving necessitate control systems capable of autonomously performing complex tasks. These control systems need to learn and adapt to unforeseen circumstances. In this PhD project, the focus lies on developing the missing theory to enable the formal verification of such autonomous control systems.   

Autonomous driving is a key application area in which theory needs to be developed to improve the verifiability and safety of autonomous control systems. Safe driving maneuvers can be expressed as dynamic constraints using temporal logics. These temporal logic specifications combine logical operations with temporal modalities based on constraints over control system's states and outputs and can be automatically verified using formal methods. However, while driving a car, inherent uncertainty is encountered that necessitates the inclusion of learning and estimation strategies. Verifying and designing such learning controllers is an open research problem for complex driving maneuvers or control tasks.

This project aims to build a framework for the safe and verifiable design of controllers for systems with state and model uncertainty. The objective of the PhD research is to exploit advanced system identification and filtering techniques (Kalman filtering, Bayesian estimation, machine learning) together with control techniques (approximate dynamic programming, Lyapunov stability, temporal logics) to improve the resilience of verifiable design methodologies to uncertainty. The research also involves some laboratory work in terms of implementation and validation on small-scale vehicles.

Main research directions

  • Investigate data-driven modelling and uncertainty quantification using system identification for Linear Time Invariant (LTI) models and Linear Parameter Varying (LPV) models.
  • Develop design methods for verifiable output-based controllers that are robust to model uncertainty.
  • Integrate abstraction techniques and model checking tools for temporal logic specification with data-driven estimation and modelling techniques.
  • Implement and apply the developed theory on suitable case studies and/or laboratory setups.

Control Systems group

The CS group research activities span all facets of systems and control theory, such as linear, nonlinear and hybrid systems theory, model predictive control, distributed control, networked systems, machine learning for control, modelling and identification, and formal methods in control. The CS group has a strong interconnection with industry via national and European funded projects in various application areas like high-precision mechatronics, power electronics, and sustainable energy (mobility, transport, smart grids). The CS group owns an Autonomous Motion Control (AMC) laboratory and hosts several high-tech setups. The PhD student will join the group and interact with the other CS group members (around 40 researchers), where he/she will participate in a mix of academic and industrial research activities. Research within the CS Group is characterized by personal supervision. The PhD student will have access to the advanced courses offered by the Dutch Institute for Systems and Control and will attend the yearly Benelux Meeting on Systems and Control.

View or Apply

Similar Positions