DC14. Road safety prediction on the basis of ethically sound physiological measurements

Aristotelis Styanidis

Education
MEng in Electrical and Computer Engineering from Aristotle University of Thessaloniki
MSc in Artificial Intelligence from KU Leuven
About
Dedicated and results-oriented electrical and computer engineer with specialized experience as a research assistant in deep learning. Demonstrated expertise in developing and implementing advanced machine learning models. Committed to fostering close collaboration with fellow engineers, emphasizing a quality-centric, end-to-end approach across projects in diverse sectors. Passionate about computer vision, signal processing, and natural language processing, with a strong focus on practical applications through machine and deep learning.

Hosts: NTUA & CARDIO

Objectives:

  • To exploit physiological measures obtained from naturalistic driving (i.e. tactile engagement of steering wheel, electrocardiogram (ECG), photoplethysmography (PPG), blood pressure, other physical activity KPIs) to create accurate and reliable real-time road safety models.
  • To investigate scenarios, involving (i) individual driving scenarios (e.g. circumventing a fixed obstacle, distraction, reverse manoeuvring) and (ii) driver interaction scenarios (e.g. overtaking, lane-changing, right-of-way negotiating).
  • To explore the ethical dimensions of driver physiological measurements in road safety assessments, the type of biases that may arise, and how these can be eliminating for more objective and fair traffic safety assessments.

Expected results:

  • Creation of innovative algorithms for physiological-based real-time prediction of surrogate safety measures with ethical considerations.
  • Production of empirical knowledge stemming from naturalistic experiments on how the aforementioned physiological KPIs vary per examined scenario.
  • Creation of assessment frameworks exploring how the ethical aspect of biometrics can be addressed in a road safety algorithmic context.

Planned secondment(s):

  • TUD-NL, Purpose: to investigate the ethical aspects of data fairness and protection in physiological measurements e.g. with respect to Regulation (EU) 2019/2144, ISO/IEC 24745:2022 – Information security, cybersecurity and privacy protection — Biometric information protection.