Can Vehicle Sensors Detect Cannabis Impairment? What the REVELIO Protocol Actually Shows
A published study protocol describes a planned test; it does not establish that the proposed detector works. The REVELIO protocol is useful for understanding how researchers intend to compare cannabis exposure, driver function, and vehicle signals, but it contains no completed diagnostic-accuracy results.
What the investigators planned
Deuber and colleagues describe a randomized, controlled, open-label pilot involving 45 healthy recreational cannabis users. The design combines baseline and subsequent driving sessions on a closed track, controlled inhaled cannabis exposure in the intervention group, and a reference group without cannabis administration.
The proposed data include vehicle controller-area-network signals, driver monitoring, physiological measures, biological sampling, and functional assessments. A dual-control vehicle and an instructor provide a controlled research setting.
Exposure markers are separate from functional references
The protocol discusses cannabinoids in capillary blood, oral fluid, and breath as exposure information. Those measurements do not become validated impairment labels merely because they are synchronized with vehicle data. The functional assessments serve a different purpose: characterizing performance and changes in driver state.
That distinction matters in machine learning. A model trained to recognize recent exposure has not necessarily been shown to recognize unsafe driving, and performance against one reference definition may differ against another.
What cannot be claimed from this publication
The protocol supplies no completed sensitivity, specificity, false-positive rate, or externally validated classification threshold. Its planned internal model evaluation cannot be cited as though it were a result. The paper explicitly describes research evaluation rather than automated restrictions or legal determinations.
Questions for future accuracy claims
Ask how impairment was defined, whether test data were independent of training data, how repeated drives from the same participant were handled, and whether the model was validated on new people and real public-road conditions. Fatigue, distraction, medical conditions, and different cannabis products are important potential challenges to specificity and generalizability.
A small controlled pilot can establish feasibility and guide later validation. This protocol provides an experimental roadmap, not evidence that an existing vehicle sensor can reliably decide whether a particular motorist is cannabis-impaired.
Related evidence
See our related analysis of toxicology evidence and its limits.
Primary research
Protocol for the REVELIO test-track pilot study: A randomised, controlled, single-centre trial in healthy recreational cannabis users investigating real-time in-vehicle detection of cannabis-impaired driving. PLOS One, 2026. DOI: 10.1371/journal.pone.0349328.
