Effects of alcohol on automated and controlled driving performances

Alcohol is the most frequently detected substance in fatal automobile crashes, but its precise mode of action is not always clear. The present study was designed to establish the influence of blood alcohol concentration as a function of the complexity of the scenarios. Road scenarios implying automatic or controlled driving performances were manipulated in order to identify which behavioral parameters were deteriorated. A single blind counterbalanced experiment was conducted on a driving simulator. Sixteen experienced drivers (25.3 +/- 2.9 years old, 8 men and 8 women) were tested with 0, 0.3, 0.5, and 0.8 g/l of alcohol. Driving scenarios varied: road tracking, car following, and an urban scenario including events inspired by real accidents. Statistical analyses were performed on driving parameters as a function of alcohol level. Automated driving parameters such as standard deviation of lateral position measured with the road tracking and car following scenarios were impaired by alcohol, notably with the highest dose. More controlled parameters such as response time to braking and number of crashes when confronted with specific events (urban scenario) were less affected by the alcohol level. Performance decrement was greater with driving scenarios involving automated processes than with scenarios involving controlled processes.