JD Discloses Embodied Intelligence Data Collection Progress: Covers Over 100 Sub-scenarios, Collects More Than 100,000 Real-world Operation Tasks
京东披露具身智能数采进展：覆盖百余细分场景、采集超 10 万真实作业任务
At the 2026 World Internet Conference Wuzhen Summit, JD Group Technical Committee Chairman Cao Peng announced that JD's embodied intelligence data collection covers over 100 sub-scenarios and has collected more than 100,000 real-world operation tasks. JD plans to collect over 10 million hours of human scenario video data and over 1 million hours of robot operation data within two years. The company highlighted its self-developed JoyEgoCam collection terminals, JoyAI-Sim physical simulation engine (with ~90% consistency between simulated and real evaluations), JoyAI-RA foundation model, and JoyScale AI computing platform. JD also outlined plans to procure 3 million robots, 1 million autonomous vehicles, and 100,000 drones over the next five years for its logistics network.
Source: Robotics — Chinese web discovery · Read original article ↗
English translation unavailable. Choose the original article to read it.
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
Reported numbers
human scenario video hours
>10,000,000 hours
View original evidence
为将场景资源转化为可训练的数据，京东正在推进具身智能数据采集中心建设，计划两年内累计采集超过 1000 万小时高质量人类真实场景视频数据，以及超过 100 万小时机器人本体运行数据，相关采集将严格依法依规进行，确保合法、合规、安全、可信。
Open source S9robot operation hours
>1,000,000 hours
View original evidence
为将场景资源转化为可训练的数据，京东正在推进具身智能数据采集中心建设，计划两年内累计采集超过 1000 万小时高质量人类真实场景视频数据，以及超过 100 万小时机器人本体运行数据，相关采集将严格依法依规进行，确保合法、合规、安全、可信。
Open source S9real task captures
100,000
robot procurement target
3,000,000 robots
autonomous vehicle procurement target
1,000,000
drone procurement target
100,000
- paper: not_reported
What remains unknown
Dataset licensing and commercial-use rights are not established here.
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
为将场景资源转化为可训练的数据，京东正在推进具身智能数据采集中心建设，计划两年内累计采集超过 1000 万小时高质量人类真实场景视频数据，以及超过 100 万小时机器人本体运行数据，相关采集将严格依法依规进行，确保合法、合规、安全、可信。
Open source S9
目前，京东具身数采已覆盖 100 余个细分生产服务场景，采集超 10 万真实作业任务。
Open source S3
例如，未来五年，京东物流计划采购 300 万台机器人、100 万台无人车和 10 万架无人机。
Open source S21
在已完成的测试中，仿真评测与真实评测结果的一致性约为 90%，评测效率达到真机评测的 3.2 倍。
Open source S13
在推理与模型服务方面，京东云自研推理引擎，大幅提升高并发场景下的吞吐与资源利用效率，典型任务推理成本最高降低 80%。
Open source S18
Implications for data suppliers
RoboSignal interpretation and collection questions, not statements of buyer demand.
- Confirm the required data type and collection setting with the buyer; this source does not establish a complete collection specification.
- Compare the reported units and scope before using these quantities in a budget. Recording hours, sensor-hours and trajectories are different measures.
- Validate demand and acceptance criteria with a buyer before scaling. Publication, popularity and a research result do not establish a purchase commitment.
Source：Robotics — Chinese web discovery · ithome.com