Record-breaking: Humanoid robot plays continuous ping pong with human for over 1000 rallies
破纪录！人形机器人与人自主连续对打乒乓球超1000拍
On November 5, 2026, a joint team from Peking University and the Beijing Institute for Foundations of AI announced that a humanoid robot based on the Zhiyuan Expy A3 platform successfully completed more than 1000 consecutive ping pong rallies with a human opponent. The match was fully autonomous without remote control or pre-set scripts, demonstrating the robot's capabilities in perception, trajectory prediction, whole-body motion planning, and ball striking under high-speed dynamic conditions. This achievement is presented as a new record in human-robot dynamic interaction, building on previous success where the same team won a ping pong championship at the HOPE World Humanoid Robot Competition using the Expy A3 platform.
Source: Chinese robotics — Hardware and sensing · Read original article ↗
Article text · Machine translation into English
Record-breaking! Humanoid robot and human autonomously play continuous table tennis for over 1000 shots
2026-11-05 22:27
Published on: Beijing
On November 5, a joint team from Peking University and Beijing Academy of Artificial Intelligence (BAAI) announced that, relying on the AGIBOT AGIBOT A3 humanoid robot platform, they successfully completed a continuous ping-pong match between a humanoid robot and a human, breaking the 1000-stroke continuous round mark and once again refreshing the new record for high-speed dynamic human-robot confrontation for humanoid robots.
This confrontation was fully autonomous and closed-loop throughout, with no remote control and no preset scripts; the robot independently completed incoming ball perception, trajectory prediction, full-body motion planning, and stroke execution, continuously responding to the variable spin, landing points, and rhythm of the human player's incoming balls, verifying the perception, decision-making, and full-body collaborative control capabilities of embodied intelligence in high-speed dynamic interaction scenarios.
Ping-pong is regarded as one of the most difficult dynamic interaction scenarios for humanoid robots, with fast ball speeds and variable spins, leaving only a millisecond-level reaction window for the robot, posing extremely high challenges to algorithms and physical control. Previously, the Peking University-BAAI joint team had used the AGIBOT A3 platform to win the ping-pong tournament at the HOPE World Humanoid Robot Sports Meet, demonstrating the powerful potential of this industry-university-research joint solution.
Industry insiders stated that this breakthrough of 1000 continuous strokes marks the transition of humanoid robots from short-term performance-style confrontation toward long-term stable and continuous dynamic interaction, further shortening the distance from the laboratory to real interaction scenarios, providing important technical verification for the deployment of humanoid robots in more dynamic work scenarios.
(Original news work by Beijing Daily News Agency)
Source: Beijing Daily Client
Reporter: Sun Qiru Return to the previous page for more information
Read ()
Platform Statement
The views in this article represent the author alone. Sohu News is an information publishing platform, and Sohu only provides information storage space services.
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- Control
- Reported autonomousOpen source S4
Reported numbers
trajectories
1,000
Trajectory count
View original evidence
人形机器人与人自主连续对打乒乓球超1000拍 2026-11-05 22:27 发布于： 北京市 11月5日，北京大学与北京智源人工智能研究院联合团队宣布，依托智元远征A3人形机器人平台，成功完成人形机器人与人类连续乒乓球对打，连续回合突破1000拍，再次刷新人形机器人高速动态人机对抗新纪录。
Open source S3
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
人形机器人与人自主连续对打乒乓球超1000拍 2026-11-05 22:27 发布于： 北京市 11月5日，北京大学与北京智源人工智能研究院联合团队宣布，依托智元远征A3人形机器人平台，成功完成人形机器人与人类连续乒乓球对打，连续回合突破1000拍，再次刷新人形机器人高速动态人机对抗新纪录。
Open source S3
本次对抗全程自主闭环，无遥控、无预设脚本，机器人独立完成来球感知、轨迹预测、全身运动规划与击球执行，持续应对人类选手多变的来球旋转、落点与节奏，验证了具身智能在高速动态交互场景下的感知、决策与全身协同控制能力。
Open source S4
此前，北大-智源联合战队就曾在HOPE世界人形机器人运动会上，使用远征A3平台拿下乒乓球赛事冠军，展现了该产学研联合方案的强大潜力 。
Open source S6
业内人士表示，此次连续1000拍的突破，标志着人形机器人从短期表演式对打，迈向长时间稳定持续动态交互，进一步缩短从实验室走向真实交互场景的距离，为人形机器人在更多动态作业场景落地提供重要技术验证。
Open source S7
（北京日报社原创新闻作品） 来源：北京日报客户端 记者：孙奇茹 返回搜狐，查看更多 阅读 ( ) 平台声明 该文观点仅代表作者本人，搜狐号系信息发布平台，搜狐仅提供信息存储空间服务。
Open source S8
破纪录！人形机器人与人自主连续对打乒乓球超1000拍
Open source S1
Source：Chinese robotics — Hardware and sensing · sohu.com