Potential for Green Deliveries with Autonomous Fleets

Recent technologies have the potential to make freight deliveries more efficient and environmentally friendly. To this end, we look at the state of the art of measuring energy consumption and how to consider this in operations research (O.R.)-based approaches for logistics planning. We also explore the potentials of autonomous fleets of robots and the role of public transit for future green deliveries.

Measuring Energy Consumption for Optimization of Conventional Fleets

Many companies are seeking to manage their environmental impact while maintaining cost competitiveness. Walmart, General Mills and Anheuser-Busch InBev all have programs designed to reduce their delivery fleets’ emissions and fuel and energy consumption. Still, minimizing costs remains one of the essential objectives for logistics. However, it is not clear how minimizing energy consumption impacts total costs. Further, most methodologies for routing problems minimize neither cost nor emissions but rather distance or travel time.

Ehmke et al. [1] explore the impact of energy and fuel consumption in conjunction with fuel and driver costs. What happens with routes and route plans when optimizing energy (fuel) consumption in urban areas versus traditional objectives? Especially in urban areas, vehicles must move at the traffic speed and are subject to the variability of those speeds at different times of the day. Next to a sophisticated vehicle routing algorithm, a detailed energy consumption model is needed to compute and compare total cost-minimized routes to those that optimize energy and fuel consumption. The authors experiment with different values for hourly driver cost, different values for fuel cost, customer geographies, customer load distributions, vehicle sizes, fleet compositions, traffic congestion, and whether the vehicle is delivering or picking up loads.

Do the additional efforts to include more detailed energy consumption models pay off? Ehmke et al. show that distance and time, objectives commonly used in routing optimization, are poor proxies for minimizing total cost and energy consumption [1]: Minimizing travel time can lead to significantly higher energy and fuel consumption than optimizing for the total costs, especially with heterogeneous fleets. Also crucial for modern businesses facing pressure to be more sustainable, the results show that minimizing total costs often increases energy consumption only minimally over routes solely optimized for energy consumption. These conclusions suggest that companies who want to minimize costs need to focus on rich cost models that incorporate the cost of drivers’ time, as well as the cost of energy consumption.

It is also an essential conclusion for routing research: Modeling total cost, particularly energy consumption, introduces non-linearities not found in traditional routing objectives. As a result, there is a need for new routing methods that overcome these challenges – both for fleets with conventional engines and electric fleets. Similar observations are also made in other papers (see, e.g., [2-3]) that explicitly consider greenhouse gas emissions in the routing models. This stream of work is mostly known under the term pollution-routing problem [4].

The Role of Public Transit for Greener Freight Deliveries

How can the flow of freight be organized in a greener way? One idea is to combine people and freight flows, which creates attractive business opportunities because the same transportation needs can be met with fewer vehicles and drivers. This can make socially desirable transport options economically viable in rural areas where the population is declining. Also, it would reduce congestion and air pollution in urban areas. It facilitates the introduction of alternative transport modes (e.g., electric vehicles). 

In the literature and practice, this line of work is denoted as “cargo hitching” aimed at more intelligent transport by better integrating people and freight modes. Given the new technology and real-time availability of information, it is possible to think ahead for unique and challenging solutions and make a significant leap forward. Clearly, as both people and goods move in the urban environment, successful integration of their streams can enhance the quality of existing transportation services and reduce congestion and pollution levels.

Actual integration is already observed in long-haul freight transportation; passenger aircraft and ferries like Norwegian Hurtigruten carry freight. Consider the following other examples: Taxis used for freight when already transporting a passenger or during idle time, coined the Share-a-Ride Problem (SARP), refers to people and parcels sharing the same taxi ride [5]. In many cities, buses travel in a fine-mazed urban network; the start and end of their tours are usually in the middle of the city. As a result, significant savings are observed, both in costs and in greenhouse gas emissions. 

Trains can replenish inventories of railway station-based stores and restaurants, which is important because railway stations are usually in time- and vehicle-restricted urban areas. Bus schedules might be adapted to accommodate the delivery of small boxes to urban retail outlets [6]. Ghilas et al. combine the freight volumes on the available public transportation operating with predetermined routes and schedules [6]. Using the scheduled lines for the freight requests shows clear benefits for the transport system. The proposed cargo hitching setup leads to significant cost reductions up to 20% using public transportation for carrying small packages. 

More recently, Mourad et al. (2021) investigated the role of under-used people-based systems to transport goods [7]. The distinguishing feature is that these authors consider a fleet of grounded and autonomous pickup and delivery robots for the first and last mile using the scheduled lines services [6]. Making effective use of these robots in an integrated system leads to significant improvements to the current system. Recently, more and more attention is given to these so-called “sideway robots” as pickup and delivery systems in the last and first mile (see, e.g., [8]).

Switching to Autonomous Fleets: Energy-minimized Operations of Robots

Because energy-efficient usage of conventional fleets seems to be more a question of algorithmic smartness and data availability, we now focus on the potential of autonomous fleets for green deliveries in urban areas. One of the most popular autonomous delivery concepts involves the use of unmanned aerial drones. These drones can carry loads of cargo over a limited distance. Drones are naturally less constrained in movement than delivery vehicles because they are flying and thus not subject to crowded traffic infrastructure. However, drone usage might become quite limited.

Unlike drones, robots autonomously driving on sidewalks to deliver a single parcel or multiple parcels seem to be a promising mode of transportation for last-mile deliveries in the same types of locations where drones may be restricted and conventional fleets are not accepted. Public opinion about these robots is generally very positive. For instance, a USPS survey says that three out of four people would accept robot deliveries. Around 30% would be willing to pay slightly more for robot delivery if the use of a robot means that the package can be delivered when and where the recipient chooses [9].

Several companies are either already selling or developing delivery robots. Starship Technologies is currently selling robots that can drive up to 6 km with a maximum speed of 6 km/h and can carry up to 10 kg in their 1.61 ft³ cargo compartments. Other companies such as Marble, Robby, FedEx and Amazon are also developing robots. For instance, Robby’s robots can go as fast as 10 km/h and can cover 35 km with one battery charge [10]. With cameras, ultrasonic obstacle detectors and GPS, a robot can operate in a highly autonomous manner with just one person supervising up to 100 of them. Starship Technologies projects that robots can decrease the cost of delivery shipments by a factor of 15, down to less than $1 for most shipments [11]. Robot delivery is already in operation in several major cities, such as London, San Francisco and Washington, D.C. [12, 13].

Bakach et al. (2021) explore local robot hubs, with robots traveling the last mile, such as at Starship Technologies in Milton Keynes, U.K. [14]. Their approach follows the idea of two-tier delivery systems where conventional trucks operate on the first tier and robots on the second tier. The idea is that human-crewed vehicles drop off many packages at second-tier local hubs. The packages would then be delivered to their final recipients by a set of robots assigned to that hub. The robots make multiple trips per day back and forth from the robot hub. Based on the p-median problem, Bakach et al. develop mathematical models for such a two-tier system that minimize the operating cost of using trucks and robots [14]. They explore different delivery options in which robots either deliver a package in a predefined customer time window or without the presence of customers.

Computational analysis shows that the cost per package for two-tiered, robot-based deliveries is much less than the conventional single-tier truck-based system across all experiments. This indicates that two-tier, robot-based delivery systems are quite promising. Robot-based deliveries can operate for about 24%-32% of the cost of conventional truck-based deliveries with sufficient customer density. This represents huge savings for delivery companies. The use of time windows increases the savings versus conventional truck-based deliveries, with savings of up to 89%. Robot-based deliveries are even more valuable with unequally distributed demand for time windows. This is when conventional deliveries cost the most per package. And finally, the energy consumption of these last-mile robots is minimal, and delivery failures can be reduced significantly.

All of these noted examples and papers show the considerable potential in cost savings and environmental benefits when integrating separated (and traditional) networks and technologies. Substantial growth and new business models are expected to increase freight e-commerce volumes and improved mode utilization (both in time and fill rate). New coordination mechanisms supported by information and communication technology solutions leading to control towers need to be designed to enable efficient integration. Additionally, price and sharing mechanisms are required to facilitate combining people and parcels.

Outlook: Use of Fleet Data to Manage Fleets in a Greener Way

The Internet of Things enables the instant exchange of data and information between machines, operators and organizations. As such, supply chains benefit from an instant exchange of information on inventory availability, supply conditions, etc. Corrective and scheduled maintenance becomes condition-based and predictable. At the same time, digital twins are used to further optimize production and maintenance processes. All these evolutions allow optimizing the performance of key operational processes, eventually leading to improvements on both internal (i.e., efficient and effective processes) and external objectives (i.e., customer value, competitive advantages). More and more, it becomes clear that data is at the heart of such decision-making; data-driven decision-making in logistics and supply chain is becoming feasible. Still, its successful implementation critically depends on the question “how?”

Much like the data-driven innovation (DDI) cycle that describes a sequence of essential phases [15], the first step is to assemble a knowledge base for demand and supply data. The knowledge base serves as a critical input for improved operational processes and decision-making in transport, logistics and supply chain management. Overall, this knowledge base contributes to better offline (proactive) and online (reactive) decision-making. As data is collected and processed into a knowledge base (keeping track of history), we have better information on stochastics, time-dependency and dynamics, improving offline decision-making (i.e., before execution). At the same time, real-time data streams enable improved online decision-making (i.e., real time during execution). For example, if something unexpected happens (e.g., accident), plans (e.g., vehicle routes) can be adapted in real time. Therefore, it is crucial to identify the relevant data that should trigger online decisions. At the same time, it is important to determine which data to include at offline decision moments. Some data might trigger online actions but might not improve offline planning.

References

1. Ehmke, J. F., Campbell, A. M. and Thomas, B. W., 2018, “Optimizing for total costs in vehicle routing in urban areas,” Transportation Research Part E: Logistics and Transportation Review, Vol. 116, pp. 242-265.
2. Jabali, O., Van Woensel, T. and de Kok, A. G., 2012, “Analysis of travel times and CO2 emissions in time-dependent vehicle routing,” Production and Operations Management, Vol. 21, No. 6, pp. 1060-1074.
3. Van Woensel, T., Creten, R. and Vandaele, N., 2001, “Managing the environmental externalities of traffic logistics: The issue of emissions,” Production and Operations Management, Vol. 10, No. 2, pp. 207-223.
4. Bektaş T. and Laporte, G., 2011, “The pollution-routing problem,” Transportation Research Part B: Methodological, Vol. 45, No. 8, pp. 1232-1250.
5. Li, B., Krushinsky, D., Reijers, H. and Van Woensel, T., 2014, “The Share-a-Ride Problem: People and parcels sharing taxis,” European Journal of Operational Research, Vol. 238, No. 1, pp. 31-40.
6. Ghilas, V., Demir, E. and Van Woensel, T., 2016, “The pickup and delivery problem with time windows and scheduled lines,” INFOR, Vol. 54, No. 2, pp. 147-167.
7. Mourad, A., Puchinger, J. and Van Woensel, T., 2021, “Integrating autonomous delivery service into a passenger transportation system,” International Journal of Production Research, Vol. 59, No. 7, pp. 2116-2139.
8. Blanco, S., 2021, “Autonomous delivery robots are now ‘pedestrians’ in Pennsylvania,” Car and Driver, March 7, https://www.caranddriver.com/news/a35756202/autonomous-delivery-robots-pedestrians-law/.
9. USPS, 2018, “Summary report: Public perception of delivery robots in the United States,” https://www.oversight.gov/sites/default/files/oig-reports/RARC-WP-18-005.pdf.
10. https://robby.io/about
11. “Starship Technologies, Inc.,” 2019, Robotics Business Review, https://www.roboticsbusinessreview.com/listing/starship-technologies/.
12. Lonsdorf, K., 2017, “Hungry? Call your neighborhood delivery robot,” NPR, March 23, https://www.npr.org/sections/alltechconsidered/2017/03/23/520848983/hungry-call-your-neighborhood-delivery-robot.
13. Teale, C., 2018, “More delivery robots could soon hit DC streets,” Supply Chain Dive, May 19, https://www.supplychaindive.com/news/delivery-robots-washington-dc-starship-last-mile/523159/.
14. Bakach, I., Campbell, A. M. and Ehmke, J. F., 2021, “A two-tier urban delivery network with robot-based deliveries,” Networks, https://doi.org/10.1002/net.22024.
15. OECD, 2015, “Data-driven innovation: Big data for growth and well-being,” Paris: OECD Publishing, http://dx.doi.org/10.1787/9789264229358-en.