Authors: Dr. Heleen Buldeo Rai, Sabrina Touami and Dr. Laetitia Dablanc
Automation promises many advantages for urban deliveries and can be classified as sidewalk robots, road robots, and drones. They employ air or road infrastructure and differ in speed, automation level, size, and carrying capacity. Regulations, public opinion, and technology costs will determine the speed at which different countries will implement autonomous vehicles for urban deliveries.
a. Advantages of vehicle automation
Automation promises many advantages for urban deliveries, including more efficient use of space, less energy consumption, improved delivery times, as well as reduced costs. As opposed to conventional vehicles, automation reduces parking requirements, particularly when drivers are manually unloading their vehicles during a delivery. This underlines the potential of reducing the space used by urban deliveries. Driverless vehicles have more capacity for the same volume, providing additional benefits in terms of load factors. With automation, fewer vehicles are needed to serve the same mobility needs, relieve congestion and its associated time delays. Automation allows vehicles a better optimization of their routing for each delivery loop and mitigates energy consumption accordingly. For urban goods transport, cost savings are expected to be quite significant due to reductions in labor costs. A Mckinsey report published in 2018 found that autonomous vehicles could reduce delivery costs in cities by approximately 10% to 40%.
A growing body of evidence underlines that autonomous vehicles are safer than conventional vehicles. They are programmed to avoid accidents and monitor all relevant context conditions autonomously and continuously to stay clear of obstacles. Removing human error, which causes about 90% of all vehicle accidents, is an additional benefit. However, some situations can rely on complex moral decisions, such as “the lesser of two evils” paradox. Experiments like MIT’s Moral Machine used crowdsourced surveys to create ethical guidelines for automation. This platform presents respondents with collision situations in which vehicles either stay in their course and hit obstacles on their path or swerve and hit something else. Accordingly, respondents were asked to judge which outcomes are more acceptable and which ones are less. Most respondents elected to sacrifice individuals to save larger groups, and most respondents spared women over men. Besides, dogs were more likely to be spared than cats, but dogs were also spared more likely than criminals. Diverging perspectives among respondents can be associated with geography and culture.
Autonomous vehicles can contribute to lower environmental emissions and noise since they rely on electrification. Autonomous delivery vehicles have the potential to reduce energy consumption and CO2 emissions by replacing internal combustion engine delivery vans and even when replacing electric vans.
b. Autonomous vehicle types for urban deliveries
Automation is not uniformly applied, but more as a sequence of improvements over five levels:
- The driver monitors the driving environment and is assisted in performing the lateral motion control (i.e. steering) or the longitudinal motion control (i.e. brake and throttle).
- The driver monitors the driving environment and is assisted in performing the lateral and longitudinal motion control.
- An automated driving system performs all dynamic tasks of driving, but the driver should be able to take control of the vehicle.
- An automated driving system performs all dynamic tasks of driving in certain conditions (e.g. highways), either occupied or unoccupied.
- An automated driving system performs all dynamic tasks of driving in all conditions, either occupied or unoccupied.
For urban goods transport, several types of autonomous vehicles are classified in a typology.
Use of infrastructure
Autonomous vehicles employ two types of infrastructure: air and road. Autonomous aerial delivery vehicles, also called “unmanned aerial vehicles” or “drones”, were initially introduced for military purposes. Because of their ability to reach difficult and remote areas, they have the potential for delivery services. Two main types are tested for delivery: multirotor drones (or “quadcopters”, “hexacopters”, “octocopters”) that are popular because of their maneuverability and hybrid drones with propellers and wings that increase their range.
Autonomous ground delivery vehicles require a well-designed and maintained land infrastructure. These vehicles can be divided into sidewalk robots and road robots. Sidewalk robots share pedestrian areas with other users. In many countries and especially in the United States, start-ups target university campuses to deploy these robots. Such environments are of interest because of their layout, but also because students are a suitable market, given that they are more smartphone adept and open to new technologies.
Depending on the type of infrastructure, the speed of autonomous delivery vehicles differs. Multirotor drones have a maximum speed of sixty kilometers per hour, while hybrid drones can go up to 120 kilometers per hour. For pedestrian safety reasons, sidewalk robots do not exceed six kilometers per hour. Road robots go much faster and have performance levels similar to regular vehicles, between 60 to 90 km/hr.
Technology allows autonomous delivery vehicles to reach their destination while avoiding obstacles through different levels of automation. Currently, drones have an automation level between three and four, meaning that they can make some decisions, but human supervision remains necessary. Semi-autonomous sidewalk robots also travel autonomously but are supervised by operators who take control if the situation warrants. They operate alongside them, at the same speed, carrying a full load of parcels. Both low-speed and high-speed road robots have reached an automation level of four, implying that they do not require human assistance in most circumstances.
Size and carrying capacity
The size and carrying capacity of autonomous vehicles are highly related. While drones carry up to five kilograms, semi-autonomous sidewalk robots generally have a capacity between 10 and 35 kilograms, up to a maximum of fifty kilograms. Such vehicles have space for one parcel, which allows them to deliver one consignee at a time. Follower sidewalk robots can carry a load of up to 300 kilograms. Low-speed road robots are smaller and lighter than regular vans but have a higher carrying capacity because they do not require a driver cabin. The design of high speed road robots is based on regular vans that are either automated or modified to accommodate autonomous deliveries. Their carrying capacity is high as well. The type of products carried by these different vehicles ranges widely, from prepared meals and groceries to regular non-food parcels of varying urgency.
c. Autonomous vehicle scenarios for urban deliveries
The use-cases in which autonomous vehicles are deployed for urban goods transport differ. Within an e-commerce context, the eight most common scenarios can be conceptualized. Each of these scenarios consists of a vehicle or combination of vehicles, as well as consumer destinations. Departure points of deliveries are either post offices, stores, or warehouses, which are either closer to or more remote from consumer destinations:
- The follower sidewalk robot scenario (1) allows delivery workers to be assisted by autonomous delivery vehicles to carry out deliveries from local post offices to consignees’ homes.
- Sidewalk robots are represented in scenarios (2) to (4). In scenario (2), the van-robot model uses vans that are specifically designed to carry a number of sidewalk robots. The van is loaded in a distribution center, drives to a specific point within a three-kilometer radius of the delivery locations, parks, and deploys the robots. Robots spread out across the radius to deliver to one consignee and return back to the van, which returns to the distribution center once robots are recollected. Labeled as a “mothership”, the van acts as a mobile hub avoiding the costs of a fixed urban distribution center.
- In scenario (3), the sidewalk robot model, sidewalk robots carry out deliveries from local businesses, such as restaurants, grocery stores, and pharmacies, directly to consumers within a three-kilometer radius. Once a consumer order is placed via a smartphone application, in-store workers prepare the order and load the robot for delivery.
- For scenario (4), the robot-cargobike model combines sidewalk robots with cargo bikes to carry out deliveries from local retail stores. In this multimodal delivery model, one cargo bike carries a few robots to a specific point close to consignees.
- Road robots are represented in scenarios (5) and (6). These robots can travel further, faster, and deliver more consignees at once. Their large carrying capacity allows storing different orders in separate compartments. These scenarios have two different starting points, from a distribution center further away from the urban area in scenario (5), the warehouse-road robot model, and local retailers, such as restaurants, grocery stores, and pharmacies in scenario (6), the store-road robot model.
- Drones are represented in scenarios (7) and (8). Scenario (7), the van-drone model, combines vans equipped with precision technology and drones. Ford Europe proposes this combination: vans drive into urban areas, and drones carry out the final deliveries to consumers.
- Finally, scenario (8), the drone model, consists of drone deliveries from local retailers such as restaurants, grocery stores, and pharmacies directly to consumers.
Although these scenarios reflect the most common and promising use of autonomous vehicles for home delivery of consumer goods in cities, new concepts are continuously planned and technological developments offer new opportunities.
d. Challenges of vehicle automation for urban deliveries
The speed at which different countries will implement autonomous vehicles for urban deliveries depends on three factors: regulations, public opinion, and costs of the applied technology. Currently, regulations do not properly address autonomous vehicles. Although drafting regulations and testing vehicles occurs simultaneously, large-scale implementation of vehicle automation still requires considerable developments of the regulatory framework. This is fundamental to ensure the fair distribution of societal advantages and to encourage more sustainable applications. Such efforts can contribute to improving public opinion, which remains largely skeptical about technological change.
Through attitudinal surveys and pilot-testing, researchers and companies try to scope expectations and perceived benefits and drawbacks of vehicle automation among the general public. For urban deliveries, particular efforts are made in the design of vehicles. Sidewalk robots that share the space with pedestrians are designed in a way that inspires a positive and nonthreatening perception by those they encounter. Yet, autonomous vehicle costs, although largely depending on the vehicle type and developer, remain high. For urban deliveries, the costs for autonomous delivery vehicles are balanced out by costs for labor. As the cost of labor increases, technology becomes relatively more affordable and accelerates the transition to vehicle automation. The question remains as if and how urban delivery concepts based on automation can create a positive business case.
Other challenges to the large-scale implementation of autonomous vehicles include adapting infrastructure and issues with liability and privacy. In terms of infrastructure, ongoing tests and implementations indicate that streets and sidewalks in some Asian and North American cities are more suitable to accommodate autonomous vehicles than European cities. To ease the testing and implementation, companies are collaborating with university campuses as well. Issues related to liability emerge because of third party involvement in the design of safety systems, causing autonomous vehicles to face greater vulnerability to lawsuits involving product liability. Should decisions be prioritized by the likelihood, severity, and quality of life effects of the type of injury, or by the number of people injured? Although it is possible to anonymize the data captured by various sensors and cameras with which autonomous vehicles are equipped, privacy issues are emerging as this provides remote surveillance opportunities by public and private agencies.
- Buldeo Rai, H., Touami, S. & Dablanc, L. (2020). Autonomous e-commerce delivery in ordinary and exceptional circumstances. The French case. Submitted to Research in Transportation Business & Management.
- Buldeo Rai, H., & Touami, S. (2020). E-commerce : vers l’automatisation du transport de marchandises en ville ? Transports Infrastructures & Mobilité, 522. http://tmv.let.fr/documents/rapports/plaquette2.pdf
- Figliozzi, M. A. (2020). Carbon emissions reductions in last mile and grocery deliveries utilizing air and ground autonomous vehicles. Transportation Research Part D: Transport and Environment, 85, 102443. https://doi.org/10.1016/j.trd.2020.102443
- Lester, C. (2019). A Study on Driverless-Car Ethics Offers a Troubling Look Into Our Values. The New Yorker. https://doi.org/10.1371/journal.pone.0001529
- Ric. (2015, September 29). How much weight can delivery drones carry? Unmanned Cargo. http://unmannedcargo.org/how-much-weight-can-delivery-drones-carry/
- Schröder, J., Heid, B., Neuhaus, F., Kässer, M., Klink, C., & Tatomir, S. (2018). Fast forwarding last-mile delivery – implications for the ecosystem.
- Society of Automotive Engineers. (2018, December 11). SAE International Releases Updated Visual Chart for Its “Levels of Driving Automation” Standard for Self-Driving Vehicles. https://www.sae.org/news/press-room/2018/12/sae-international-releases-updated-visual-chart-for-its-“levels-of-driving-automation”-standard-for-self-driving-vehicles
- Steer. (2020). Economic Impacts of Autonomous Delivery Services in the US.
- Touami, S. (2020). Les robots de livraison en ville, une solution à venir ? Université Gustave Eiffel, chaire Logistics City.