Garbage Collection in Chicago: A Dynamic Scheduling Model
研究芝加哥垃圾车调度，利用马尔可夫决策过程设计灵活路线，根据街区垃圾产生率调整每日倾倒次数，在五个试点区可减少12-16%的卡车容量。
We investigate the scheduling of garbage trucks in the city of Chicago. Analysis of data collected from the system shows that city blocks differ in the rate at which garbage is collected. However, in the current system, each truck visits the dumpsite two times each day. Our approach is to devise a flexible routing scheme in which some routes visit the dumpsite only once per day, while others visit the dumpsite twice per day depending on the blocks assigned to the route. We use a Markov decision process to model the impact on capacity of using flexible routes. This provides a dynamic scheduling algorithm that adjusts the number of dumpsite visits throughout the week to maximize service level. Results of the model suggest a potential reduction in truck capacity of 12–16% for a set of five pilot wards. This paper shows that flexible schedules can significantly reduce the capacity required to operate a system in the presence of variability.