Conference Proceeding
Authors: Nicholas J Kelling (Georgia Institute of Technology, Atlanta) , Lana M Corso (Georgia Institute of Technology, Atlanta)
Driving has become an integral part of our daily lives, and so too have the dangers associated with driving. Understanding driver behavior could lead to system modifications to alleviate some of these inherent dangers. Specifically, prediction of driver braking behavior might be used to improve automatic braking systems and adaptive cruise control systems. The research presented in this paper details the development of an algorithm to predict the brake onset times in situations where rear-end collisions might occur. The algorithm is adaptive to an individual and not set to generic values. This algorithm was generated using data from a previous study (Kelling, 2006). Displayed stimuli consisted of different situations for a lead vehicle (stopped, slower moving, and reversing lead vehicle), multiple rates of closure (32.2, 64.4, and 96.6 kph), and two luminance conditions (day or night driving). Brake onset times were recorded. A self-modifiable algorithm was developed and was found to have an R-squared value of .625. The degree of goodness-of-fit for this algorithm is worthy of note because it also considers differences in the driving environment. The individualized adaptive ability of the algorithm provides a greater overall fit for predicting braking behavior, and it may be more useful in automated systems than existing algorithms.
Keywords:
How to Cite: Kelling, N. & Corso, L. (2007) “Prediction of Brake Onset Times for Rear End Collisions”, Driving Assessment Conference. 4(2007). doi: https://doi.org/10.17077/drivingassessment.1239