Tech giant Alibaba Group is deploying a fleet of robots to fulfil the demand for online shopping across China.

The company is deploying 1000 delivery bots across Chinese university campuses and urban communities this year. The bots, called Xiaomanlv or “small donkey” in Mandarin, can deliver about 50 packages at a time and as many as 500 boxes in one day, covering 100km on a single charge.

The bots pick up parcels at a local courier drop-off point and make their way to the customer’s building, trundling along sidewalks and in bicycle lanes. They can’t navigate a flight of stairs yet, but the company says that technology is steadily improving.

In the future, the bots could gradually replace forklifts in factories, dispose of medical waste in hospitals and even move luggage around airports.

The idea is these bots will learn how to navigate obstacles safely in low-speed minor roads and cul-de-sacs, such as serving a college campus or within a housing estate. The firm’s Damo Academy will analyse insights gathered from these early models and then upgrade the bots to be safe to manoeuvre even on public roads.

“In three to five years, we hope to make progress in autonomous driving in faster speed scenarios and make deliveries over longer distances,” said Wang Gang, head of Damo’s autonomous driving lab in Hangzhou.

Last-kilometre delivery has always been the knotty problem of ecommerce. It’s costly, time-consuming and largely unmapped. This final leg in a journey that stretches from the manufacturer to the buyer can flummox human couriers. Delivery people can get lost trying to find a flat in a tower block or navigate a large housing estate.

Alibaba’s robots will be able to make the trip without deviations or smoke breaks. An algorithm plots the fastest route for the robot to the door. 

The robot can predict the intended movement of passengers and vehicles up to five to ten seconds in the future and safely avoid collisions. In test runs, 99.9999% of the time, the Xiaomanlv’s auto-driving algorithm manages without human intervention. More importantly, the robot’s positioning technology allows it to operate even where there’s weak or no GPS signal.

China’s ecommerce market is the largest in the world and its growth has stimulated the swift development of logistics to fulfil orders. The coronavirus pandemic has also accelerated the shift in consumption from offline to online and heightened demand for contactless delivery of packages.

Package deliveries topped 830 billion in mainland China last year, nine times the volume of 2013, according to China’s Post Office. At the same time, there are fewer able-bodied couriers in China, one of the fastest aging populations globally. The World Health Organisation expects the percentage of people over 60 years old in China to hit 28% by 2040.

“We do not have enough labour to deliver so many packages to consumers,” said Wang Gang. “It’s impossible without automated technologies. So, this is why Xiaomanlv is very important.”

Wang Gang has spent over a decade researching artificial intelligence as an academic and now wants to put theory into practice.

“I hope to build AI products that can have practical application and a big social impact, rather than academic impact,” he said. “Alibaba’s Damo Academy’s objective is to develop products that can be used, so this aligns well with my motivation.”

As consumers demand faster and more flexible delivery schedules, ecommerce companies are racing to rewire their logistics networks. The Hangzhou-headquartered group offers customers in China one-hour delivery, half-day delivery and next-day delivery.

“Nowadays, consumers have higher expectations such as delivery to their building so that they do not have to walk for miles to get their package,” said Wang.

Damo Academy is working closely with Alibaba’s logistics arm Cainiao to deploy the autonomous mobile robots. Delivery companies such as FedEx and UPS still handle most packages globally. However, a few retailers are exploring more cutting-edge technology such as drone deliveries. That said, there are relatively few drones in circulation, given the high level of regulatory clearance needed. Delivery robots, or bots, on the other hand hold more promise in terms of mass deployment.

Alibaba’s approach is to build steadily on the insights gleaned from the bots already in action, increasing the likelihood of developing a more mature and useful algorithm rather than relying on engineers in the lab.

Wang believes Alibaba’s algorithm is strong enough not to need expensive, high-definition sensors that some companies’ prototypes need to navigate.

“So, we can lean more on our algorithm to achieve a low-cost, mass deployment of the robots,” he said.