Seminal
The magazine and atlas of EU-funded researchA new publication from the web and AI team at Scott Charnley

Atlas / Austria

Marie Skłodowska-Curie Action · Running

RHive-ML Reliable High-emitting Vehicle identification using Machine Learning

Coordinator
Technische Universitaet Graz, Austria
Runs
1 September 2027 to 31 August 2029
Status
Running
EU funding
€230k
Total cost
—
Partners
2 organisations in 2 countries
Type
Marie Skłodowska-Curie Action
Grant agreement
101282591

What the project is doing

In the project's own words, from its summary on CORDIS.

Air pollution remains a leading environmental health risk, with road traffic a major contributor. Although strict regulations have reduced emissions from new vehicles, two persistent challenges undermine progress: a minority of high emitters caused by malfunctioning or tampered aftertreatment systems, and the rising contribution of non-exhaust emissions from brakes, tyres, and road dust. Point Sampling (PS) provides a cost-effective, accurate approach to capture real-world emissions, but its use is constrained by the need to distinguish combustion-based from non-exhaust particles and by reliance on number plate data, which raises privacy concerns.

This project will overcome these limitations by separating combustion from non-exhaust emissions through correlation of particle metrics (PN, BC) with CO2, where strong correlation signals combustion-related sources. Based on this separation, vehicles that only emit non-exhaust particles can be identified as electric vehicles (EVs). Machine learning (ML) models will then be developed on combustion-based emissions to infer fuel type and emission standard without number plate readings, and to translate PS snapshots into representative average emissions to identify high emitters while distinguishing them from Diesel Particulate Filter regeneration events.

The models will be trained on detailed emission profiles derived from chassis dynamometer and Portable Emission Measurement System data and applied to PS datasets from European field campaigns, with robustness and interpretability ensured through state-of-the-art ML algorithms and interpretable ML methods. The outcome will be a privacy-respecting, PS-based framework for large-scale, real-time detection of high emitters, enabling stronger enforcement of emission standards, effective mitigation of traffic-related air pollution, reliable identification of EVs from non-exhaust-only signatures, and future-proofing of European emission control policies.

Topics

Who is involved

OrganisationCityCountryRole
Technische Universitaet GrazGrazAustriaCoordinator
Technische Universitaet DresdenDresdenGermanyAssociatedpartner

Source: CORDIS, European Commission. Data of 5 Nov 2026, reused under the Commission's reuse policy. View on CORDIS · Grant DOI