Intelligent Algorithms for Sustainable Transportation

20 September, 2021
Algorithms for sustainable transportation


Article by Majsa Ammouriova, Leandro do Carmo and Rafael Tordecilla, researchers of the IN3’s ICSO research group.


Transportation and logistics systems represent a key element in global economics and energy consumption, and play an essential role in distributing goods across supply chain networks. These systems include heterogeneous fleets of traditional internal combustion engine vehicles and other types of vehicles using “greener” technologies, such as bicycles, electric vehicles, hybrid vehicles and even unmanned or self-driving vehicles.

Transportation planning and logistics systems face various challenges, some being tied to the main activities carried out in these systems, such as vehicle routing and scheduling. Further challenges have also presented themselves, firstly because of increased global awareness of environmental and sustainability issues and, secondly, due to concern over health issues caused by high NOx concentrations and excessive noise exposure caused by urban road traffic.

Specifically, these challenges include bringing greener vehicles into transportation and logistics, minimizing the carbon footprint of goods distribution in urban areas, increasing the number of electric vehicle charging stations available, and bolstering infrastructure for bicycles and other types of vehicles. There is a rising need to both address these challenges and propose solutions for sustainable development. 


Global changes call for advanced algorithms in transportation and logistics

The fast-growing Internet of Things and advancements in information and communication technology have led to the emergence of new business models, such as e-commerce. These models increase competition and create new challenges for transportation and logistics systems.

New information technology makes it possible to collect, store and process big data in cities in real time, giving rise to so-called smart cities, whose scope combines sustainable development with intelligent data management. As a result, services such as smart waste management, ride-sharing and carpooling are operated more efficiently, and charging stations and facilities are placed in optimal locations. 

Global changes call for algorithms that are able to account for environmental and sustainability factors, overcome the challenges posed by big data and recommend solutions for optimizing the efficiency of transportation and logistics systems. Computational algorithms have evolved to deal with these challenges.

Analytics algorithms, for instance, can perform many functions: describe and track key performance indicators related to transportation and logistics through descriptive analytics, define patterns and causes of key performance indicators’ behavior through diagnostic analytics, forecast these indicators through predictive analytics, and support decision-making and compare different possible scenarios through prospective analytics. 


Algorithms combining simulation and optimization to overcome stochasticity

Advanced algorithms combining simulation and optimization have been widely used to overcome stochasticity in transportation and logistics systems. Recently, the term x-heuristics has been used to define algorithms, with examples including metaheuristics, matheuristics, simheuristics, biased-randomized heuristics and learnheuristics.

These algorithms vary in their application depending on their characteristics. While metaheuristic algorithms are general-purpose heuristics that may solve any number of optimization problems, simheuristics integrates simulation with optimization to encompass stochasticity and uncertainty simultaneously. Finally, learnheuristics combines machine learning with metaheuristics to cope with dynamic problems. 


Agile optimization algorithms: proposing efficient solutions in real time

As smart cities evolve, the need for algorithms capable of handling real-time data and big data associated with large systems grows ever greater. Accordingly, agile optimization algorithms play an increasingly important role in solving dynamic problems in large systems and proposing efficient solutions in real time.

These agile algorithms are based on the massive parallelization of biased-randomized heuristics, which are extremely fast in execution, flexible and adaptable to different problems, parameterless, and suitable for online optimization. This means that agile algorithms utilize available data for recommending solutions even if these data are incomplete, with optimization being resumed when new data are collected. For instance, the arrival of new orders means updating the order log, arranging activities and rescheduling vehicles. Likewise, unexpected traffic jams can easily affect route planning and scheduling. In such cases, these algorithms can support the agile selection of solutions while considering the smart city’s sustainable development. 


2021 SI-Trans Workshop for Young Researchers

Advanced algorithms are fundamental tools for supporting strategic, tactical and operational planning in transportation and logistics systems, ensuring efficient and sustainable activities in urban areas. Because of their impact, researchers have turned their focus on these algorithms and their applications.

One event tailored to highlight the use of intelligent algorithms was the 2021 SI-Trans Workshop for Young Researchers organized by the IN3’s Internet Computing & Systems Optimization (ICSO) research group, which is part of the Spanish R&D Network in Sustainable and Intelligent Transportation, supported by the Spanish Ministry of Science, Innovation and Universities, and the Erasmus+ SEPIE. At the event, which took place on 12 and 13 July 2021, young researchers as well as some senior participants presented their studies in transportation and logistics and highlighted recent trends in these fields.

The topics included the use of unmanned aerial vehicles (drones) in transportation, the use of time series forecasting to predict freight companies’ transportation loads, and the measurement of economic losses and recovery in air transportation. The participating researchers also discussed problems related to sustainability and the COVID-19 pandemic.

The scope of the workshop was not limited to simply discussing problems facing transportation and logistics systems. Researchers were also welcome to showcase any tools they had devised to deal with these problems or solutions focusing on sustainability, namely analytical tools, machine learning and simheuristics. The research underlined the need to employ different types of advanced and agile algorithms, in order to efficiently cope with real-world transportation problems in smart cities, which are ever more dynamic, stochastic and complex, and require increasingly sustainable alternative solutions.


Article by Majsa Ammouriova (postdoctoral researcher), Leandro do Carmo and Rafael Tordecilla (PhD students) are members of the Internet Computing & Systems Optimization (ICSO) research group at the Internet Interdisciplinary Institute (IN3).

The ICSO group focuses on the use of intelligent algorithms and data science (including optimization, simulation, analytics and machine learning methods) to support complex decision-making in different fields of application, including transportation and logistics, smart cities, production, real-time positioning and computational finance.