Chinese Robot Plays Tennis Pro: Embodied AI Breaks Physical Limits
新智元 · wechat · 2026-08-22
Galactic General's robot played the first-ever human-vs-robot tennis match against pro player Zheng Jie, marking a breakthrough for embodied AI in dynamic adversarial scenarios. Tennis demands perception, decision-making, and whole-body control within milliseconds, serving as an extreme stress test.
Technical Highlights:
- Unified Architecture: The "AstraBrain" integrates high-level decision-making (cerebrum) and motion control (cerebellum) into a single model. This eliminates latency and information loss found in traditional layered architectures.
- Training Pipeline: Combines "imperfect human data" with simulated self-evolution. The model learns priors from human motions for a cold start, then engages in millions of simulated matches. This leads to skill emergence (e.g., diving saves, self-recovery) without explicit hard-coding.
Performance: The robot demonstrated tactical positioning in doubles and handled high-speed serves, spins, and slices in singles. Notably, it autonomously stood up after a fall and continued playing, validating the closed-loop capability from cognitive decision to physical execution.
Related event: Galactic Robots Play First Human-Machine Tennis Match(3 posts)→
More from Embodied
- MIT's C-GMT gives humanoid robots whole-body compliance without sacrificing motion tracking — MIT_CSAIL · 2026-10-16
- OpenRoboto, Bittensor's largest robotics subnet, puts its data collectors on Solana via peaqOS — bittingthembits · 2026-10-16
- Robotics data tokenization landscape: Khala maps peaq, THOR, BitRobot and Axis — 0xSammy · 2026-10-16
- Perseverance rover logs 28.1 miles on Mars, breaking Opportunity's record in half the time — BrettKrieger12 · 2026-10-16
- Trossen bimanual arms portion food in commercial kitchen with models fine-tuned via Physical Intelligence API — kscottz · 2026-10-16
- Safeworld emerges from stealth with $12M to build 'CI/CD for robots' safety testing — lukas_m_ziegler · 2026-10-16