JD.com's "Wolf Pack" robots make a collective debut at the world's largest scale: ZhiWolf, FlyingWolf, and LoneWolf.

At the 2026 World Robot Conference held in Beijing's Yizhuang district, JD Logistics showcased its fully self-developed "Wolf Pack" robot军团, marking the largest AI-driven supply chain robot fleet globally, covering the entire chain from warehousing, picking, and transportation to delivery.

Three main robots were exhibited on-site: ZhiWolf, FeiWolf, and DuWolf.

ZhiWolf is a goods-to-person system that works alongside vertical lifts and transport robots, delivering containers to pickers in as fast as one minute. It transforms the traditional person-to-goods model into goods-to-person, boosting warehouse picking efficiency.

FeiWolf is a logistics drone supporting three operation modes: airdrop, sling loading, and belly cargo. With over 100 drone routes already operational in China, it can complete deliveries in mountainous areas within as little as three minutes.

DuWolf is an L4-level autonomous delivery vehicle capable of assessing road conditions and avoiding obstacles on its own. Nighttime delivery routes using it have already been launched in Shenzhen.

The entire robot fleet is coordinated by JD Logistics' self-developed SuperBrain large model 2.0. This model is specifically designed for supply chain scenarios, not a general-purpose one, and can complete billion-package simulation reasoning in three minutes while handling over 1,000 real-world business scenarios.

When order surges outpace capacity, the system automatically adjusts inventory and allocates tasks between humans and machines. This smart logistics project has also been selected as a UN Global Digital Economy Lighthouse Case, earning international recognition.

Beyond the "Wolf Pack" robots, JD also brought a robot maintenance service station called "CyberMed." It handles the full maintenance cycle—fault diagnosis, battery swapping, performance testing, and equipment recycling—and this service has already expanded into European and Middle Eastern markets.

In short, what JD showcased this time isn't just a few robot hardware units, but a complete end-to-end solution from scheduling and operations to after-sales maintenance. For large numbers of robots to run reliably in real-world settings, hardware strength alone isn't enough—backend large-model scheduling and a comprehensive repair and maintenance system are equally critical.