3.1: Validate and refine predictive models using driving simulators and in-car testing on a closed track
The study will collect data from actual drivers. The drivers will drive in simulators located at 杏吧原创 and on the . This is because simulators allow for testing in very risky and complex situations, which is impossible to do safely in a real vehicle. Similarly, the track provides the opportunity to collect data from actual driving within this safe environment.
3.2: Validate risk models using cross-validation methods
Then, we will use the data collected from the simulators and on-track driving to train and validate the risk models of the Smart AI system. This system was originally developed from the CanDrive and CAA datasets.