HARMO

Enhancing road safety for automated vehicles through a better understanding of real driver behaviour

Automated vehicles can only be safe in real traffic if they can predict what human drivers and the other road users will do. In cities, drivers on familiar routes often rely on habit rather than constant vigilance, yet most safety assessments still assume an alert, textbook driver.

HARMO (Human-Aware Road Mobility Optimisation) is a three-year Horizon Europe project that aims to enhance road safety for connected, cooperative and automated mobility (CCAM). It builds validated models of how route familiarity changes human driving behaviour. These models help automated vehicles anticipate other road users, and behave in ways others can anticipate in return. They also give developers, test facilities and regulators a common, evidence-based way to judge whether that behaviour is safe.

HARMO builds on the EU projects i4Driving and BERTHA. It combines human behavioural data with a hybrid AI planner, tests it in realistic mixed-traffic simulations, and benchmarks it in urban logistics and shuttle pilots in shadow mode, where it observes and predicts without controlling the vehicle.

Why this project matters to IRU

Safe automated driving depends on predictable interaction between vehicles and the people around them, including professional drivers, pedestrians and cyclists.
Clear, harmonised safety evidence gives regulators a sound basis for decisions and gives operators more certainty before they invest in automated solutions.
IRU makes sure the operator and professional driver perspective is built into safety assessment, regulation and standardisation.

IRU's role

IRU leads HARMO’s work on acceptance, regulatory alignment and market uptake, bringing the perspective of road transport operators and professional drivers to the question of what safety evidence automated vehicles must provide before they are trusted on European roads.

Safety acceptance: IRU gathers the expectations of regulators, operators, insurers and social partners, and uses its knowledge of professional driving to ground the pass/fail criteria for automated vehicles.
Regulatory alignment: IRU maps how EU, UNECE and national rules treat the validation of automated driving behaviour, and leads the Safety Evidence Pack and Regulatory Evidence Toolkit to support type approval and trial admission.
Policy and uptake: IRU leads the project’s policy engagement, including at least two policy briefs, and shares HARMO’s results with operators and public authorities through its networks, events, intelligence reports and UNECE engagement.

Project results

HARMO is expected to deliver validated models of familiarity-driven human behaviour at technology readiness level 5, together with a hybrid AI planner that produces human-like, explainable and verifiably safe driving. It will also produce assessment metrics aligned with ISO 26262, ISO/PAS 21448 and UNECE rules for automated driving, and traceable evidence packages for CCAM safety claims.

An Acceptance Threshold Matrix will set measurable thresholds for trust, perceived safety and predictability, tested with drivers, vulnerable road users and passengers. The project will also provide a Regulatory Evidence Toolkit, Human-Predictability Guidance and an adoption roadmap, and will share open, GDPR-compliant datasets and scenarios through European CCAM validation platforms.

Project type:

EU-funded innovation and research

Start date:

1 Oct 2026

End date:

30 Sep 2029

Partners:

ITS Norway coordinates the project. The other partners are FZI Forschungszentrum Informatik, Transport & Mobility Leuven, Nord University, LOXO, ISFM Intelligent Systems for Mobility - member of MILLA group, ENIDE Solutions, Fundación CIDAUT, the University of Surrey, SafeAD, Trygg Trafikk and CTAG.

Project Manager:

Silvia Barbaro

Contact