What are emerging technologies for urban and last-mile deliveries?
How can optimization and AI-powered analytics help reduce the pollution & congestion caused by freight transportation in urban areas?
E-commerce has experienced exponential growth in the last few years. Recent statistics have shown that, while retail e-commerce sales were around 1 billion dollars in 2014, they are estimated to reach 5 billion dollars in 2021. While this represents a huge business opportunity for companies in almost all fields, it also raises dramatic issues in terms of managing the operations related to satisfying customer orders on time. In particular, focusing on the last leg of the supply chain, the e-commerce explosion has put a spotlight on last-mile delivery. In fact, customers are becoming more and more demanding in terms of delivery speed. On one hand, offering services like ‘same-day delivery’ or ‘delivery within the next day’ represents a prime opportunity for increasing revenues and obtaining customer loyalty. On the other hand, this means reducing consolidation opportunities and time for delivery planning.
The main consequence is a massive increase in the number of commercial vehicles going around road networks for deliveries: given the short delivery time requested by customers, parcels are sent out as soon as they are available, with little to no consolidation. This means having a lot of almost empty vans on roads, which leads to pollution, congestion, and deterioration of quality of life. A big portion of this traffic is condensed in urban areas, and, consequently, the negative outcomes explode in this context. These booming e-commerce activities (representing, 8.4% of annual growth in France, source: https://www.statista.com/outlook/dmo/ecommerce/france), along with the increasing demand to reduce CO2 emissions within city boundaries, are inevitably inducing a paradigm shift in the daily operations of logistic service providers (LSPs). LSPs are looking for innovative strategies and business models for improving the status quo and making last-mile deliveries environmentally friendly and sustainable.
Stakeholders are starting to develop and propose solutions for this huge issue . They range from sharing strategies,like crowdshipping and freight deliveries on public transit systems, to green distribution means, like electric vehicles, robots and drones.
Whatever is the means to tackle the issue, optimization and AI-powered analytics can help, and might even be crucial, in reaching the goal through the development of technologies enabling the proper management of the tools envisaged. How does it work? We describe it here.
The “sharing economy” is a term that identifies emerging activities, services and initiatives, whereby people and organizations share their available resources with potential users to obtain a mutual benefit. This also happens in city logistics where ordinary people, i.e., not professional drivers, offer their time and resources to provide transportation services. This phenomenon is called “crowdshipping”. One prominent example is Amazon Flex, which was introduced in 2013 and, nowadays, is widely used in the US and starting to be used in Europe. Crowdshipping is also associated with the term uberization which means that ordinary people make available their time and resources (car, fuel, ...) to other transportation services (either for transporting people or freight). Indeed, Uber has launched two projects associated with freight transportation: Uber Freight and Uber Eats.
Why is crowdshipping becoming so popular? The main reason is that it is cost-effective: indeed, companies can cut down fixed costs related to hiring and wages, while paying just for the service provided. It thus represents a great business opportunity. However, this does not come for free: the organization of the distribution process becomes much more complex when dealing with crowdsourced drivers that reveal their availability shortly before the service needs. As a consequence, an optimized distribution plan needs to be foreseen in order to avoid wasting the benefits coming from fixed-cost savings. Optimization technologies represent the right tool: while taking into account all demand and service requirements, they can build the most effective distribution plan (, ).
Figure 3. Crowdshipping
Freight on Transit (FOT)
In 2014, freight deliveries accounted to 15% of urban traffic. With the explosion of same-day delivery service in the last few years, we can expect this statistic to have increased dramatically. As the foundational piece in building a smart city, city logistics plays an important role in reducing the fossil fuels consumption caused by freight transportation.
In our Sustainable Smart City Operations (SISCO) research project funded by the CY Initiative of Excellence , we propose leveraging existing public transportation services during off-peak hours, when the vehicles are typically under-utilized, to help LSPs deliver packages in the urban areas. This novel logistics concept of integrating goods and passenger flows to promote higher utilization rates for the public transport network is known as Freight on Transit (FOT) [2,3]. In FOT, public transport operators cover the “first leg” of transportation which is then combined with green modalities performing the “last leg”. In a promising FOT pilot project launched by Monoprix in Paris (2007-2017), the company used the RER line D to transport goods between their distribution center in Combs-la-Ville and the city boundary of Paris (Bercy) . A fleet of around 20 natural gas trucks was then used to perform the last-mile delivery, and distribute goods from Bercy to around 60 Monoprix shops inParis. According to the company’s estimations, this produced remarkable environmental gains: annual CO2 emissions have been reduced by 280 tons, and around 10,000 truck deliveries (on an annual basis) have been replaced by the train.
The Monoprix project showcases great potential for other LSPs in implementing a similar FOT concept. Besides the two major stakeholders (i.e., the public transport operators and LSPs), the last-mile delivery part may also involve the third party logistics providers operating with drones/robots, micro-logistics operators, or individuals (crowdshipping). Due to its complexity and the traditional organizational structures, the adoption of the FOT may be a challenging task for the two major stakeholders. To overcome these barriers, the goal of our project is to provide decision making tools which can be used to estimate expected environmental impacts, and to answer important strategic, tactical or operational questions, such as which lines should be used for freight transportation, which stations should be the entry and exit points, the size of the required fleet, and how to route the packages for the last-mile delivery. These important managerial insights will help the decision-makers make informed decisions based on data, optimization and analytics.
Last-mile delivery with low-emission vehicles, drones and robots
In 2016, the cost of global parcel delivery, excluding pickup, line-haul, and sorting, totalling approximately 70 billion euros. According to the McKinsey report , over the next ten years, market volumes in Germany and the US might reach 5 billion and 25 billion parcels per year, respectively. The biggest share (often higher than 50%) in total parcel delivery cost goes to last-mile delivery. This is why the large and highly dynamic parcel delivery market is constantly being disrupted. Innovative last-mile concepts have been proposed to cope with the increasing demand for logistic efficiency and competitive prices. Among them, one can now find pickup points networks, integrated public and freight transportation, deliveries directly into the customer’s c trunk, crowdshipping, and more recently, the use of unmanned aerial vehicles (drones) and self-driving autonomous robots.
From a regulation point of view, the adoption of drones has been rendered increasingly difficult around the world due to the adoption of stricter rules concerning their operation and safety, especially in urban areas. In this context, self-driving robots have an advantage as they are designed to operate at low speeds, e.g., pedestrian speed, so that they can safely share existing sidewalks and bike lanes with people. Self-driving delivery robots were introduced much later than drones, nevertheless, many initiatives can now be found where robots are deployed for deliveries. For example, the self-driving robots developed by e-novia (2020), Starship (2020), and Twinswheel (2020) have been tested in many cities around the world. More recently, Amazon also announced the development of their own self-driving delivery robots, called Scout (Amazon, 2020). FedEx tested a six-wheeled, autonomous robot, called the SameDay Bot, in summer 2019.
In this context, several operational decision and network design problems arise. One is the selection of robot stations for the last-mile delivery of parcels via robots and the optimal routing of the truck transporting parcels to these selected stations from a central depot (Alfandari, Ljubic, Melo da Silva, 2020) [6,7]. This paper proposes mathematical models and methods to optimize Quality of Service (which means to minimize tardiness relatively to customers’ due dates). Given the complexity and large problem size (tens of potential robot stations and hundreds of customers), the paper proposes efficient methods (namely, Benders decomposition) that can find optimal strategies and enable to explore insightful what-if scenarios (with respect to impact of robot speed, robot range, network structure) that are relevant for practitioners and companies. For example, increasing the speed of robots from 5 km/h to 15 km/h, results in annual savings of 675 kg CO2, for a single urban area represented by a 10 km square grid considered in our study. For the given instance the truck route is reduced by more than 50%, whereas the average distance traveled by robots increases by 45%, and fewer facilities are visited. Increasing the coverage radius of robots from 30 to 60 minutes has the highest environmental impact with 750 kg annual CO2 emissions savings for the 10 km square grid considered in our study.
 SISCO - Sustainable Smart City Operations: Research project funded by the CY Initiative of Excellence (grant "Investissements d'Avenir" ANR-16-IDEX-0008). Website: https://sites.google.com/a/essec.edu/sisco/home
 Ozturk Onur: Freight on Transit as a new concept for city logistics, https://aqtr.com/association/actualites/freight-transit-new-concept-city-logistics
 Keith Cochrane, Shoshanna Saxe, Matthew J. Roorda & Amer Shalaby (2017) Moving freight on public transit: Best practices, challenges, and opportunities, International Journal of Sustainable Transportation, 11:2, 120-132, DOI: 10.1080/15568318.2016.1197349
 Antoine Boudet: Monoprix choisit Fret SNCF pour approvisionner ses magasins parisiens, Les Echos, 5 juil. 2007, https://www.lesechos.fr/2007/07/monoprix-choisit-fret-sncf-pour-approvisionner-ses-magasins-parisiens-534433
 Joerss, M., Schroder, J., Neuhaus, F., Klink, C., & Mann, F. (2016). Parcel delivery: The future of last mile. McKinsey&Company, . McKinsey Report on Travel, Transport and Logistics.
 Alfandari L., Ljubic, I., Melo da Silva, M. (2019). Optimal Vehicle Routing with autonomous devices for last-mile delivery. In: 2019 Workshop of the EURO Working Group on Vehicle Routing and Logistics optimization (CeRoLog 2019),
 Alfandari L., Ljubic, I., Melo da Silva, M. (2021). A tailored Benders decomposition approach for last-mile delivery with autonomous robots, Optimization Online, http://www.optimization-online.org/DB_HTML/2021/03/8279.html, European Journal of Operational Research (to appear).
 Archetti, C. , Bertazzi, L. (2021). Recent challenges in routing and inventory routing: e-commerce and last-mile delivery, Networks 77, 255-268.
 Archetti, C., Savelsbergh, M., Speranza, M.G. (2016). The vehicle routing problem with occasional drivers. European Journal of Operational Research 254, 472-480.
 Archetti, C., Guerriero, F., Macrina, G. (2021). The online vehicle routing problem with occasional drivers. Computers and Operations Research, 127, 105144.