The HCSM behavior models in Hank provide a convenient means to orchestrate complex scenarios required by experimenters interested in studying human driving performance. In order to compare performance across subject groups, the essential aspects of events and situations must be repeated from trial to trial. This must be accomplished without perceptible contrivance in the face of unpredictable variation in driver behavior.
In our research, we use HCSM director objects to create the critical situations and events that compose an scenario [6]. These directors are pure behavior objects (that is they have no dynamic or visual model.) They monitor the simulation through database queries and send directives to vehicles, traffic lights, and pedestrian objects. This leads to a style of control that is opportunistic and improvisational. By this we mean that the roles objects play in a scenario and their detailed behaviors are determined online. For example, to create an intersection incursion, an appropriate vehicle is selected to be the culprit as the driver nears the intersection. Because each trial evolves differently, different vehicles will be selected as opportunities arise.