Apple Discloses HALO Humanoid Robot Control System; Unitree G1 Achieves 91% Average Success Rate on Single Tasks
苹果披露HALO人形机器人控制系统 宇树G1实测单项任务成功率91%
Apple's research team submitted a paper disclosing the HALO autonomous control system for humanoid robots, validated on Unitree G1 in real-world environments. The system demonstrated actions such as pushing boxes, following people, sitting on chairs, and navigating obstacles. In six unscanned or unmapped environments, 360 single-task tests yielded an average success rate of 91%. In one environment, three combination tasks averaged a 70% success rate over 10 trials each, with a specific sequence achieving 50%.
Source: Chinese robotics — Hardware and sensing · Read original article ↗
Article text · Machine translation into English
The Apple research team submitted a paper disclosing the humanoid robot autonomous control system HALO, and deployed it on Unitree G1 to complete real-world environment verification: in 6 real environments that were not pre-scanned or mapped, the average success rate for 360 single-item task tests was 91%.
Viewpoint Network News: Recently, the Apple research team submitted a paper disclosing the humanoid robot autonomous control system HALO, and verified it by deploying it on the Unitree G1 humanoid robot, allowing the robot to apply actions learned in simulation to unfamiliar real environments. The team released data generation, model training, and deployment code, demonstrating actions such as pushing blocks, following a person, sitting in a chair, and crossing obstacles.
In 6 real environments that were not pre-scanned or mapped, the average success rate for 360 single-item task tests was 91%.
In another real environment, 10 tests were conducted for each of the 3 combination tasks, with an average success rate of 70%, among which the “bend down to pass obstacle—walk towards target—push box” combination had a success rate of 50%.
Disclaimer: The content and data in this article are compiled by Viewpoint based on public information and do not constitute investment advice; please verify before use.
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- Control
- Reported autonomousOpen source S2
Reported numbers
Reported success rate
91%
Summed sensor-hours
View original evidence
苹果研究团队提交论文披露人形机器人自主控制系统HALO，并部署在宇树科技G1上完成真实环境验证：6个未预先扫描或建图的真实环境中，360次单项任务测试平均成功率为91%。
Open source S2test_count
360
Reported trials
View original evidence
苹果研究团队提交论文披露人形机器人自主控制系统HALO，并部署在宇树科技G1上完成真实环境验证：6个未预先扫描或建图的真实环境中，360次单项任务测试平均成功率为91%。
Open source S2environment_count
6
View original evidence
苹果研究团队提交论文披露人形机器人自主控制系统HALO，并部署在宇树科技G1上完成真实环境验证：6个未预先扫描或建图的真实环境中，360次单项任务测试平均成功率为91%。
Open source S2Reported success rate
70%
Summed sensor-hours
View original evidence
另在一个真实环境中，对3种组合任务分别测试10次，平均成功率为70%，其中“俯身通过障碍—走向目标—推箱子”组合成功率为50%。
Open source S6test_count
10
Reported trials
View original evidence
另在一个真实环境中，对3种组合任务分别测试10次，平均成功率为70%，其中“俯身通过障碍—走向目标—推箱子”组合成功率为50%。
Open source S6Reported success rate
50%
Summed sensor-hours
View original evidence
另在一个真实环境中，对3种组合任务分别测试10次，平均成功率为70%，其中“俯身通过障碍—走向目标—推箱子”组合成功率为50%。
Open source S6
- paper: not_reported
- code: Reported pending Open source S4
- dataset: not_reported
- weights: not_reported
- code: not_reported
What remains unknown
Not established in the collected evidence: Environment, Data origin.
Reported performance applies to the described task. It does not establish general autonomy or deployment readiness.
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
苹果研究团队提交论文披露人形机器人自主控制系统HALO，并部署在宇树科技G1上完成真实环境验证：6个未预先扫描或建图的真实环境中，360次单项任务测试平均成功率为91%。
Open source S2
另在一个真实环境中，对3种组合任务分别测试10次，平均成功率为70%，其中“俯身通过障碍—走向目标—推箱子”组合成功率为50%。
Open source S6
团队公开了数据生成、模型训练和部署代码，展示推箱子、跟随人、坐椅子及穿越障碍等动作。
Open source S4
Source：Chinese robotics — Hardware and sensing · guandian.cn