By capturing millimeter-precision human movement data, JD.com is building a proprietary dataset to train a new generation of robots, a move that could give it a significant edge in the race for embodied artificial intelligence.
By capturing millimeter-precision human movement data, JD.com is building a proprietary dataset to train a new generation of robots, a move that could give it a significant edge in the race for embodied artificial intelligence.

(P1) Chinese e-commerce giant JD.com Inc. has launched a data collection initiative in Suqian, China, to train its embodied AI systems, using a custom-designed 220-gram wearable device to capture human movements. The project aims to create a unique, high-precision dataset that could accelerate the development of more capable logistics and service robots, challenging competitors in a field that relies heavily on simulated data.
(P2) The announcement, made in partnership with the Suqian municipal government, details the creation of China's first "embodied intelligence data collection community." According to the company, the program will capture the nuances of human interaction with the physical world, gathering data that is difficult or impossible to replicate in virtual environments.
(P3) At the heart of the project is the JoyEgoCam, a 220-gram, head-worn terminal developed by JD.com. The device is equipped with an automotive-grade inertial measurement unit (IMU) that enables millimeter-level precision in tracking upper limb trajectories, force distribution, and environmental interactions during tasks like cleaning, folding clothes, and organizing objects.
(P4) This investment in bespoke data collection is a critical component of JD's plan to increase its AI-related R&D spending by over 200% this year. By creating a rich library of real-world interaction data, the company aims to improve the performance of its logistics and automation platforms, potentially unlocking significant operational efficiencies and creating a defensible moat for its AI models, including the JoyAI and JoyIndustrial LLMs.
The Suqian initiative is a concrete step in JD.com's broader strategy to build what it calls the world's largest "Physical AI operations hub." The company is moving to integrate AI into more than 3,000 use cases across its retail, logistics, and healthcare operations. During its recent 618 Grand Promotion, JD heavily promoted AI-enabled products, from smart mattresses to cooking robots, powered by its JoyInside AI agent.
The company expects JoyInside to be integrated into more than 10 million smart hardware devices this year, spanning nearly 200 brands. The data collected by the JoyEgoCam will directly feed the development of these and future embodied AI systems, which are AIs designed to interact with and manipulate the physical world. This real-world data could prove to be a significant competitive advantage over rivals who primarily rely on synthetic data generated in simulations, which often fails to capture the complexity of real-world physics and human behavior.
However, the project also raises questions about data privacy, a growing concern for consumers and regulators worldwide. While JD.com's project focuses on capturing movement and interaction data, any large-scale collection of personal information brings potential risks. A recent report from the Electronic Privacy Information Center (EPIC) found that many major technology companies use manipulative designs to make it difficult for users to opt out of data selling and sharing.
The EPIC report highlights a broader industry trend of opaque data practices, noting that even when companies claim not to sell data, they often share it with partners for advertising and other purposes. For consumers, the ability to control how their data is collected and used is becoming increasingly difficult. While JD.com's initiative is for a specific research purpose, it operates within a tech landscape where the lines around data ownership, consent, and use are frequently blurred, putting the onus on companies to establish clear and transparent privacy protocols.
This article is for informational purposes only and does not constitute investment advice.