Learning from the whole spectrum of driver behaviour: from unsafe to optimal driving

Research objectives

• Understand and learn from safety as a spectrum, based on the Safety-II concept.
• Identify optimal driving behaviours in different environments and contexts.
• Analyse the contributing factors and collective impact on the safety of different driver behaviours.
• Derive insights for ADAS applications and real-world safety interventions.

⚙Research progress

I am currently working with large-scale naturalistic driving datasets to identify and analyse driver adaptation patterns, with a focus on car-following behaviour. My work includes the development of a methodology to detect adaptation episodes based on relative velocity dynamics, and the application of data-driven methods such as DTW-based clustering to uncover temporal behaviour patterns. These analyses provide insights into how drivers adapt to different traffic situations and what characterises safe, unsafe, and optimal driving.