GravityControl: Natural Language Correction for Embodied Intelligence
颠覆重训模式！极智嘉GravityControl上线：自然语言一喊就纠偏，具身智能终于‘听得懂人话’了
JiZhiJia has launched GravityControl, the world's first real-time natural language correction model for embodied intelligence, enabling robots to adjust their paths in real-time during shelf navigation through natural language commands, without requiring algorithm engineers or additional training data, increasing task success rate from 17.5% to 87.5%.
GravityControl represents a significant shift in embodied intelligence by enabling real-time natural language correction without retraining, addressing long-standing challenges in real-world adaptability and operational efficiency.
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
Reported success rate
17.5%
View original evidence
测试数据显示，面对从未见过的物体姿态、突发光照变化、传送带轻微偏移等典型‘分布外（OOD）’长尾场景，任务整体成功率从惨淡的17.5%飙升至87.5%，单点操作如精密插拔、柔性抓取的成功率更是从5%狂飙到95%。
Open source S7Reported success rate
87.5%
View original evidence
测试数据显示，面对从未见过的物体姿态、突发光照变化、传送带轻微偏移等典型‘分布外（OOD）’长尾场景，任务整体成功率从惨淡的17.5%飙升至87.5%，单点操作如精密插拔、柔性抓取的成功率更是从5%狂飙到95%。
Open source S7Reported success rate
5%
View original evidence
测试数据显示，面对从未见过的物体姿态、突发光照变化、传送带轻微偏移等典型‘分布外（OOD）’长尾场景，任务整体成功率从惨淡的17.5%飙升至87.5%，单点操作如精密插拔、柔性抓取的成功率更是从5%狂飙到95%。
Open source S7Reported success rate
95%
View original evidence
测试数据显示，面对从未见过的物体姿态、突发光照变化、传送带轻微偏移等典型‘分布外（OOD）’长尾场景，任务整体成功率从惨淡的17.5%飙升至87.5%，单点操作如精密插拔、柔性抓取的成功率更是从5%狂飙到95%。
Open source S7robot trajectories
12,000
View original evidence
极智嘉中报披露，其具身产品已拿下宝洁、丰田供应链、京东亚洲一号等十余家世界500强订单，预计2026–2028三年累计出货超1.2万台，场景正从‘仓内搬运’加速向‘产线装配’‘零售补货’‘医院物流’破壁延伸。
Open source S18
- paper: not_reported
What remains unknown
Not established in the collected evidence: Environment, Control, 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.
测试数据显示，面对从未见过的物体姿态、突发光照变化、传送带轻微偏移等典型‘分布外（OOD）’长尾场景，任务整体成功率从惨淡的17.5%飙升至87.5%，单点操作如精密插拔、柔性抓取的成功率更是从5%狂飙到95%。
Open source S7
极智嘉手握全球市占率前三的AMR底盘，每年数百万小时的实车运行数据，早已把仓库变成了全球最大规模的具身模型实景训练场。
Open source S16
极智嘉中报披露，其具身产品已拿下宝洁、丰田供应链、京东亚洲一号等十余家世界500强订单，预计2026–2028三年累计出货超1.2万台，场景正从‘仓内搬运’加速向‘产线装配’‘零售补货’‘医院物流’破壁延伸。
Open source S18
Source：Robotics — Chinese web discovery · sohu.com