Could autonomous vehicles put last-mile delivery on the fast track?

By Peng He and Jialin Li, World Economic Forum

Transport, integral to our daily lives, is currently undergoing a structural reform – towards digitalization and decarbonization. The COVID pandemic has accelerated the shifts, with everyone adopting new norms. But while social distancing and travel restrictions have meant most people are much less itinerant, other transport sectors are experiencing huge growth. In the parcel delivery sector, the number of items distributed worldwide increased by 17.5%, in 2020 reaching a new high of nearly 25 billion items.

Meanwhile, as the virtual reality and metaverse concepts continue to hint at what people might one do be able to do without stepping out of their homes, e-commerce and online shopping continue to grow rapidly. Since the pandemic started, sales grew an additional 27.6% up until the end of 2020, accounting for 18% of the total global retail market.

Urban last-mile delivery (LMD), a core aspect of the package delivery value chain, is the end leg of an item’s journey, from a transportation hub to a final destination. The most labour-intensive stage of delivery, scattered customer distribution and the high frequency of requests means it accounts for a high proportion of costs across the wider logistics chain. Purpose-designed shopping festivals, like Double 11 in China, and Black Friday and Cyber Monday in western countries, during which time period the number of parcels drastically increases, are a major driver behind the rapid development of this means of urban delivery.

The challenge that LMD is currently facing goes far beyond just building an effective and efficient sorting system. Perennial inner-city problems, such as traffic congestion and carbon emissions, as well as the double parking and the increasing number of delivery vehicles, are becoming ever more disruptive issue for urban transport systems – which naturally affects LMD. It is also an activity heavily reliant on human labour, with its demand for workforce fluctuating depending on peak and idle hours of business.

Autonomous vehicles (AVs) could be a solution. Their on-board technology, such as intelligent driving and electric powertrain, means traffic congestion and carbon emissions can be significantly reduced. AVs could also serve to supplement to human delivery to diversify services, and to fill workforce shortages during the busy periods and night hours.

Number of parcels distributed worldwide from 2015 to 2020, by sector (in million units)
Number of parcels distributed worldwide from 2015 to 2020, by sector (in million units)
Image: Statista

Europe, China, and the United States are the regions that have been pioneering the trend of automotive electrification, with electric vehicle (EV) penetration rate in new sales of 10%, 5.7% and 2% in 2020 respectively. In the field of autonomous last-mile delivery, the US is taking the lead, with major players like Starship Technologies and Nuro in the market. In China, trial runs have been conducted only not in logistic parks, private communities and university campuses, but also on open public roads in megacities like Beijing and Shenzhen. Other cities around the world, such as Guangzhou, are taking steps to pioneer this new solution.

Certain ALMDV characteristics make them a brilliant testing platform for cutting-edge intelligent driving technologies. Compared with passenger AVs, ALMDVs are smaller and are normally restricted to relatively low travel speeds, which limits the risk of accidents. With no humans on board, algorithms are generally focused on protecting pedestrians on public roads, without the decision-making dilemma of prioritizing passenger or pedestrian in emergent situations. Besides the technology architecture of the vehicle itself, such as sensoring, positioning and operating, it also provides a good opportunity to apply and test the V2X (vehicle-to-everything) concept. One of the core technological accomplishments of the Fourth Industrial Revolution, this governs how smart vehicles communicate with roadside infrastructure such as traffic lights, including the cloud operation system.

There are also undoubtedly challenges on the demand side. Today’s customers expect immediate or same-day deliveries. Speed of delivery accounts for 52% of decision-making for purchases from customers worldwide, a more important factor than free or discounted shipping fees (38%). As the society has adopted social distancing, more parcels are delivered to a central hub, like parcel lockers, where customers come and pick them up; AV delivery would allow more parcels to delivered right to the doorstep, within a predictable time window, while minimizing human contact.

The application of intelligent driving on a small, flexible device creates a host of scenarios that will require imaginative solutions. Like how ALMDVs can get up stairs to deliver parcels to doorsteps. One idea is to place another autonomous device at the entrance, like a housekeeper robot, with access to the building’s elevator system to take the parcel from the ALMDV and carry it onwards. A new retailing model could be incubated through ALMDVs: one with not only the capability to deliver goods from a nearby store, but that also can be used as a remote displaying and selling device; a “store on the move”, in communities, campuses and tourism areas.

To make ALMDVs a daily reality, the first step is legislation. There are various ways to categorize AVs: people-carriers or goods-carriers; operating on public roads or private property; high speed or low speed, and so on. But which type of regulations should be apply to the Autonomous Last Mile Delivery Vehicle (ALMDV)? Is it a vehicle, a non-motor vehicle, a personal delivery device, or a robot? The answer to this question ultimately determines which lane an ALMDV will be allowed to drive. On roadways, it is allowed to travel at faster speeds, and the lanes are shared with other vehicles rather than pedestrians. On sidewalks and off roads, the speed limits are usually much lower, but there are more potentially dangerous interactions with people. Discussions about how to define ALMDVs are ongoing among different stakeholders, with the hope of establishing a new category for them.

At the turning point of decarbonization and digitalization of the automotive and transport industries, we face both opportunities and challenges. Intelligent driving technologies offer new business models, user scenarios and lifestyles. The industry will benefit from collaboration across different sectors, such as between policy-makers and enterprises. Innovation and technological breakthroughs need a supportive environment of flexible policies and regulations. It’s exciting to see ALMDV pilot projects spring up in different regions, experiences being shared between the pioneers and those following in their wake, and more discussion between different stakeholders. The mass adoption of intelligent driving is around the corner now – let’s move it, and society, into the next gear.