Tangent point oriented curve negotiation

Drivers are known to frequently fixate on the tangent point during curve negotiation. The question is how they use the tangent point to obtain the input signal(s) necessary to steer their car through a curve. In this paper, an answer is provided in terms of a steering control model that uses the distance to the tangent point and the car's heading relative to the tangent point vector. These two quantities are intermittently registered and used to update the steering angle. Experimental data obtained in our fixed based driving simulator shows good agreement with the model. They show similar behavior in terms of anticipation (steer into curve before the road starts to bend) as well as cutting through the inner side of the curve to keep the maximum curvature of the car's trajectory as low as possible. Besides providing insight into human steering behavior and a framework in which many hypothesis about curve negotiation can be tested, one of the applications for this model lies in the area of automatic steering control.