Theoretical and experimental investigations were conducted of automobile path deviations when a driver is attempting to steer his vehicle along a straight path with his vision occluded. A three-factor (car, subject, speed), two-level field experiment was carried out to test for main and interaction effects. Another field experiment was carried out to determine the effects of no steering control. In both experiments, the vehicle path deviations from the theoretical straight path were measured over several hundred feet and were used as the dependent variable. Both experiments were conducted on a flat airport runway under daylight and no-windconditions. The experimental results indicate no significant car or subject effects but a highly significant speed effect and a significant car-subject interaction. Specifically, the standard deviations of the vehicle displacements from the theoretical straight path are considerably smaller at the higher speed for a given distance traveled under occluded vision. Further, the standard deviations of vehicle displacements for a given distance traveled are considerably larger for the no steering control condition than for the steering control condition with no visual input. The experimentally obtained data seem in basic agreement with the theoretical path deviation model. Based on the experimental data, the distribution of vehicle displacements for a given distance traveled under no visual input could be reasonably approximated by a normal distribution.

• AN UNCERTAINTY MODEL in which the driver is treated as an information processing device has been developed (4, 5). The analyses were based on steady-state driving in which the driver's vision-was intermittently occluded, and on the assumption that a driver's uncertainty between two consecutive looks stems from (a) the loss of relevant road information ahead of him because he forgot or it became obsolete and (b) his uncertainty about the vehicle's lateral position on the road because of random disturbances in the orientation of the vehicle. This study, discussed in more detail elsewhere (1), deals only with vehicle position uncertainty and demonstrates that the functional relationship as derived (4, 5) cannot be supported by experimental evidence. A new functional relationship that is in better agreement with the experimental data is developed in this study. Theoretical studies dealing with driver steering control have been conducted by a number of authors (6) and are the background for the theoretical development of the new functional relationship.