Dick de Waard and Karel A. Brookhuis
Centre for Environmental and Traffic Psychology University of Groningen, The Netherlands
From manual control to full automation five different grades of automation can be discerned ( Endsley & Kiris, 1995). Starting from (1) no automation, via (2) decision support, (3) consensual Artificial Intelligence (AI), (4) monitored AI to (5) full automation, authority of the human being decreases, while the role of the system increases. In (1) a (cognitive) task is accomplished without any assistance, while in (2) automation provides advice. In (3) the system proposes actions and the operator's consent is required while in (4) the operator can only veto actions initiated by the system. In full automation the operator can and does not interact with the automated system at all (see also Endsley, 1996). Similar to aviation, automation in surface travel ranges from manual control to completely automated driving. Most systems that are low in level of automation do not restrict the driver's behavior. A system such as RDS-TMC (Radio Data System - Traffic Message Channel) provides the driver with information on congestion. Such a system is an indirect route-guidance system, leaving the decision to take another route to the driver. Automation is at the level of "decision support", according to Endsley's classification. At the other extreme there is automated driving, such as in the Automated Highway System (AHS, e.g., Congress, 1994). In the AHS cars exchange information with the road infrastructure and with each other, enabling automatic lateral (steering) and longitudinal (speed and headway) control. Only when cars are in the AHS or on dedicated lanes, they will be under automatic control. Advantages of AHS include increased road capacity, increased operation in bad weather and reduction or even elimination of driver error. The latter claim is based on the assumption that drivers will not be allowed to take over control while driving on the AHS, which is still a point of discussion ( Tsao et al., 1993). As one of the aims of AHS is to increase road capacity, cars will probably drive at short following distances, at time headways that are beyond human reaction time. AHS is likely to leave very little, if any, behavioral freedom in actual driving (i.e., in headway and speed control), and is the highest level of automation.
Michon ( 1985) has described driving as a task with processes at three levels. At the top level, the strategic level, strategic decisions are made, such as the choice of means of transport, setting of a route goal, and route-choice while driving. At the intermediate level, the maneuvering level, reactions to local situations including reactions to the behavior of other traffic participants take place. At the lowest level, the control level, the basic vehicle-control processes occur, such as lateral-position control. Automation in driving can take place at all three levels. The earlier mentioned RDS-TMC provides the driver with strategic information, while the AHS has full operational control. At the tactical level information or tutoring systems can inform drivers about their behavior. These tutoring systems can range from silent black-box policing systems to environmental feedback systems. In Figure I some of these systems are categorized according to the level of automation and the level of process that are automated.
According to Wickens ( 1992) there are 3 goals of automation, each serving a different purpose. There is 1) automation to perform tasks humans cannot perform at all, or 2) automation to perform tasks