2019 Industrial AI Sketch: Logistics Edition
-
The highly structured nature of logistics work makes AI applications particularly effective, especially in transportation and delivery, which have seen partial implementation. However, due to the high R&D costs of AI hardware, commercial development of logistics AI still has a way to go.
In the Industrial AI series, we've introduced a range of AI applications that made significant progress in 2019. Yet, many might still feel it's not enough. Whether it's AI in the cloud reducing costs and improving efficiency for the financial sector or AI monitoring crops in the fields, these technologies are subtly transforming our lives. However, they still fall short of the revolutionary changes we envision.
Admittedly, our expectations for technology are often exaggerated and dramatic—like associating biotechnology with biohazards or imagining AI applications that completely overhaul ancient industries, drastically improve efficiency, and even introduce robots with extraordinary skills. Isn't that far more 'exciting' than merely relying on cloud-based algorithms to transform industries?
Such sci-fi-like scenarios actually emerged in 2019, particularly in AI-powered logistics.
When people think of logistics, a montage likely plays in their minds—packages being tossed around in massive warehouses, illegible shipping labels, and delivery riders braving the cold to rush through cities. Such an industry seems resistant to technological transformation, remaining heavily reliant on manual labor, at least until autonomous vehicles become mainstream.
However, while consumers mostly interact with delivery personnel, logistics is actually ripe for AI integration.
First, from a motivational standpoint, the stereotypes about logistics—such as inefficiencies in warehousing and difficulties in information sorting—are real challenges the industry faces. There's clear willingness to adopt technology for improvement.
Additionally, labor shortages in delivery networks have become increasingly evident over the past two years. With parcel lockers gradually replacing door-to-door deliveries, the industry is eager to leverage technology to address this 'labor crunch.'
Despite its seemingly 'primitive' appearance, logistics work is highly structured. For instance, item placement in warehouses and worker routes are often fixed, as are the distribution processes for logistics vehicles across cities.
Logistics heavily relies on navigation systems, generating vast amounts of data from daily operations. Moreover, the industry is inherently clustered and scalable, making its workflows ideal for structured processing and AI-driven optimization. Even seemingly distant advancements like autonomous driving have made significant progress in fixed-route scenarios, naturally aligning with logistics hubs.
For years, logistics has never lacked intelligence. Large logistics companies often showcase smart sorting lines and automated warehousing systems where robotic arms and conveyor belts handle packages without human intervention.
In 2019, the most notable change in logistics AI was the decoupling of business and technology. While large logistics firms have high internal adoption rates for smart solutions, these technologies are often tightly bound to specific business needs, limiting the emergence of mature, standalone AI solution providers. This has hindered industry-wide technological adoption, creating a gap between flashy PR demonstrations and the reality of overworked delivery personnel.
This year, however, more tech players—from giants like Huawei and Baidu to smaller firms—have begun addressing logistics challenges.
The most visible AI advancements in 2019 were in transportation. For example, data mining combined with AI algorithms can optimize delivery routes and improve mileage efficiency. Tasks like scheduling, which once took skilled workers hours, can now be completed by AI in minutes. Historical data analysis also enables freight demand forecasting, allowing sites to prepare in advance.
Take shopping festivals like Singles' Day or 618. Without proper planning, warehouses risk overloading, exhausting workers, and creating hazardous stockpiles. Overestimating demand, however, leads to unnecessary costs. AI helps strike a balance, easing peak-period pressures. Meanwhile, warehousing AI has also improved.
Autonomous vehicles are increasingly deployed in ports for cargo transfer, while flexible, small-scale warehouse robots—enhanced by digital twin technology—are becoming more common. Logistics firms no longer need massive smart factories; smaller robots can collaborate to boost sorting efficiency and reduce mishandling.
Finally, in the delivery phase, 'magical realism' scenarios are unfolding. Since last year, companies like Suning and Amazon have tested autonomous delivery vehicles for the 'last mile.' In 2019, such solutions proliferated—Deppon's unmanned vehicles appeared on university campuses, and China Post's autonomous helpers aided delivery personnel during Singles' Day.
In short, the future we imagined is materializing.
However, while logistics AI is feasible, its commercialization remains distant. Replacing delivery personnel with autonomous vehicles might cut labor costs, but the steep price tag—around ¥800,000 per unit—deters most firms. Logistics thrives on scale, but scaling such technology requires massive investment. Equipping hundreds of urban logistics sites with autonomous vehicles could mean nearly ¥100 million in costs.
Looking ahead to 2020, two trends are likely: First, hardware costs will decline due to competition and R&D. Products like warehouse robots and autonomous delivery vehicles are still in their early adoption phase, and prices may drop as subsidies and competition intensify, broadening accessibility.
Second, easier-to-deploy software solutions will integrate deeper into logistics. Facial recognition for parcel lockers or OCR for document processing are low-cost AI applications that enhance user experience and serve as entry points for broader adoption.
As labor shortages affect various sectors, logistics AI can be repurposed—for food delivery, port scheduling, and more. Both logistics and tech firms will likely increase investments here, possibly leading to logistics companies pivoting into tech service providers.
In the convergence of AI and the physical world, logistics will undoubtedly be a standout chapter—but patience is required.