Driver Impairment Multimodal Evaluation (DIME)
| Reference number | |
| Coordinator | Göteborgs universitet - Göteborgs universitet IT-fakulteten |
| Funding from Vinnova | SEK 3 000 000 |
| Project duration | August 2025 - August 2027 |
| Status | Ongoing |
| Venture | Safe automated driving – FFI |
| Call | Traffic-safe automation - FFI - spring 2025 |
Purpose and goal
The Driver Impairment Multimodal Evaluation (DIME) project addresses the challenge of identifying intoxication, particularly when drivers actively attempt to conceal it through creating an advanced multi-modal system, designed for commercial deployment by 2028. For this, Smart Eye will leverage deep learning techniques to develop their Driver Monitoring System to combine: i) in-cabin driver behavior; ii) intoxication-related speech patterns; and iii) biobehavioral measures.
Expected effects and result
DIME is expected to produce: 1) an advanced multimodal (in-cabin - bodily posture/face, speech recognition - including interactive language content, biobehaviour - including heart-rate, eye movement, steering) algorithm for detecting cognitive impairment in drivers, specifically alcohol intoxication, and 2) integrate the algorithm into a driver monitoring system demonstrated in simulated environments. The aim will be to reduce within 6-10 years intoxication-related driver accidents by 20%.
Planned approach and implementation
The project will involve two steps of modelling: i) in-cabin behaviour modelling of intoxicated drivers, ii) fusion (fused model) of a) in-cabin behaviour model, b) intoxicated speech detection model, c) biobehavioural model. Furthermore, modelling will be enabled by analysis of existing and new data. New data will be obtained using Smart Eye’s driver simulator. Controlled simulation studies will test biobehavioural, in-cabin behaviour and speech/linguistic markers of intoxication.