A dynamic data-driven model for optimizing waste collection
Authors
P. Sarvari, I.A. Ikhelef, S. Faye, and D. Khadraoui
Reference
in the 2020 IEEE Symposium Series on Computational Intelligence, SSCI 2020, Virtual, Canberra, Australia, 1-4 December, art. no. 9308221, pp. 1958-1967, ISBN: 978-172812547-3, 2020
Description
Commercial waste collection activities are critical from environmental, societal, as well as economic perspectives. Logistic activities carried out in any large or small human settlement, must be efficient by passing through obstacles while optimizing rare and valuable resource usages. With the advent of the Internet of Things and smart waste management ideas, the concept of static waste collection resource optimization and more specifically vehicle routing problem are being exposed to a fortunate mutation. This study introduces a dynamic waste collection optimization model and its solution for a unique type of waste collection problem. Unlike public waste collection, which is made up of homogeneous customers, commercial waste collection has to consider other factors, relating to the quality or time of service, while considering the socio-economic characteristics of the customers. Moreover, the paper has completed a comprehensive literature review over the waste collection filed to emphasize the singularity of the problem and the proposed mathematical model. The data-driven model proposed in this paper targets the optimization of costs in the embedded solver with invoking real-time data generated by filllevel sensors integrated into waste containers. The outputs of the model are dynamic and time-wise vehicle routing chains for efficient waste collection under the field official guidelines, constraints, and priorities. In order to scrutinize the scalability, applicability and validity of the proposed model, a real-life network in Luxembourg with multiple vehicles, stops, as well as a depot and a disposal site has been considered. The partnership with a waste management company, called Polygone, benchmarking results with real data conclude the merits, excellence, and findings of the paper.
Link
doi:10.1109/SSCI47803.2020.9308221
