TREE NEWS update: Apple’s research team submitted a paper titled “Perceptive Humanoid Motion from Simulation” to the arXiv preprint platform on November 4, disclosing an autonomous humanoid robot control system called HALO. The system was deployed on Unitree Robotics’ G1 humanoid robot to validate that motions learned in simulation could be executed in unfamiliar real-world rooms, achieving an average success rate of 91%. The work calls on the Qwen model in testing.
Apple’s HALO Humanoid Control System Hits 91% Success Rate on Unitree G1
Apple's entry into humanoid control research, validated on a third-party Unitree platform rather than its own hardware, signals that the sim-to-real gap is becoming a software problem that large tech firms can attack with foundation models. The 91% success rate in unfamiliar rooms matters less as a benchmark than as evidence that simulation-trained policies can generalize without per-room tuning. The open question is whether this remains a research preprint or becomes a licensing or partnership thread between Apple and robotics vendors like Unitree.
Generated by AI for reference only.
Share on WeChat
Open WeChat → Scan → then tap "…" to send to a chat or Moments.
Tap "…" in the top-right corner to send to a chat or share to Moments.