decreases the risk of possible damage
or harm when an accident happens.
Finally, a delivery vehicle is free of
passengers, therefore, it has fewer requirements for safety, planning, and
There are two major challenges in
last-mile delivery, which autonomous
driving can help: The distribution location of the package, and the delivery path. From past experience, when
a delivery person arrived on location,
the majority of time was spent waiting,
especially when the planned delivery
consisted of a large number of small
packages destined for office buildings, campuses, and apartments. The
waiting time for customers, and the
handling time to delivery to customers, killed any efficiency of last-mile
delivery. Moreover, the delivery person
is paid by the number of packages delivered, meaning the company often
pays a great deal of money for a trip to a
single location, which kills any cost efficiency of last-mile delivery.
To be effective, last-mile delivery
must determine the best route to dispense the most parcels. In a city, the
best route is often not the shortest
route, and road conditions constantly
changed over time. An autonomous
driving cart is similar to a larger “
self-closing cabinet,” which can save both
the average waiting time and the distribution time. On the other hand, equipment costs increase with the autonomous approach. Ideally, there would
be a cost of only ¥ 1. 5 per autonomous
delivery compared with current ¥ 7–10
per delivery. Achieving this requires
reducing costs of trial carts from
¥600,000 to ¥ 50,000 if the autonomous
delivery (and vehicles) are in mass production.
Autonomous driving vehicles have
major technology challenges, too.
One obstacle is the behavior of the
motion detection system when out
in the real world. The algorithm may
work perfectly in lab test conditions,
but may not perform well when it is on
the open road. Another difficulty with
these vehicles is the range of vision
when driving in dark or shadowing areas. Like the human eye, the range of
the vision may vary in different levels
of brightness; even with infrared light
detection, the vehicle may not “see” in
foggy and dusted environments.
How JD Uses the Vehicles
in Delivery Scenarios
JD.com is the largest e-commerce platform by revenue, and offers a world-class set of online retail services to its
legion of users, who now number close
to 300 million in total. As a technology-driven company, JD.com has focused
considerable effort in developing a robust and scalable retail platform that
not only supports the company’s rapid
growth but also allows it to provide cutting-edge technology and services to its
partners and customers.
JD selected last-mile delivery prac-
tices as the first line of defense in its
campaign to upgrade its logistics infra-
structure. JD Logistics’ autonomous de-
livery vehicles will primarily be used for
last-mile delivery in urban areas, carry-
ing packages from dedicated stations to
office buildings, pick-up stations, resi-
dential area convenience stores, and
other locations. It was first used to sup-
port JD.com’s renowned two-hour ex-
press delivery service and will be rolled
out across JD’s deliver network to be
used for a wider range of applications.
Autonomous driving vehicles will be
loaded at delivery stations and will travel
to pick-up points designated in advance
by consumers. The recipient can collect
the products they order simply by press-
ing a button on JD’s mobile app. JD’s
delivery vehicle can also recognize the
customer using face identification and
deliver the product accordingly.
JD conducted its first trial in autonomous driving vehicles for last-mile delivery on June 18, 2017 at Renmin University, Beijing. The vehicle delivered
about 10 packages in approximately
six hours. JD subsequently deployed
approximately 60 autonomous driving
vehicles for last-mile delivery at Beijing, Xian, and Hangzhou for pilot AI-based package delivery. The city of Xian
has been selected as the headquarters
for JD’s fleet of vehicles. In December
2017, JD Group CEO Qiangdong Liu announced last-mile parcel delivery plans
for 100 universities.
Figure 3. The deployment progress of JD’s autonomous driving vehicle.
Figure 4. The use cases of the JD autonomous driving vehicle.