Specification and development of scheduling algorithms for managing driver fatigue within road transport

Abstract

Fatigue (sleepiness, tiredness) is the largest identifi-able and preventable cause of accidents in transport operations (between 15% and 20% of all accidents), more than that of alcohol or drug related incidents in all modes of transportation. Drowsiness can produce the same types of effects as alcohol; slowed reaction time, decreased awareness and impaired judgment, all of which increase the risk of having an accident. Driver fatigue should be considered in the early phases of planning and making of driving schedules. Drive and rest regulations are imposed by different councils to ensure a marginal fatigue management. This thesis concerns development of an algorithm which makes optimized driving schedules to manage drivers' fatigue for commercial transports. Here the vehicle routing problem (VRP) with time windows was solved using dynamic programming. Then an alertness prediction model, the sleep-wake predictor, was used to evaluate the schedules in the first step and then was incorporated into the algorithm to produce optimized schedules based on predicted alertness. Alertness based optimization improved the average driving alertness, and also decreased the duration of high accident risk (HRD) intervals. In the simulations, the alertness level has increased 4% and HRD decreases by 9% in average. However, as a result, the traveled distance will go up 22%, which will increase the total cost in another way. A more detailed analysis can be done considering actual accident and regular transportation costs, and further improve the algorithm to be used in real life scheduling problems.