Same policy, same task, different bodies.同一策略,同一任务,不同本体。
A policy runs read-only beside a human operator, and we measure how far it drifts, and how that changes when you swap the arm or hand.策略在人类操作员旁只读运行,我们测它偏离人多远,以及换臂或换手之后偏离怎么变。
Leave your details留下联系方式
Demo: nothing was sent.演示:什么都没有发出。
In production this goes to the ERP lead intake (contact@ → lead_intake worker). This is exactly what that message would contain:正式上线后,这会进入 ERP 线索收件(contact@ → lead_intake worker)。那条消息的内容就是下面这些:
What it can and can't say today它今天能说什么、不能说什么
No real recordings yet. The booth replays synthetic rehearsal data, so no "body X vs body Y" claim can be made from it.还没有真实录制。展台回放的是合成彩排数据,所以不能据此得出任何「本体 X 对比本体 Y」的结论。
i
What the instrument does today这台仪器今天能做什么
- Records human + policy side by side under one clock and one file format.在同一时钟、同一文件格式下并排记录人和策略。
- Computes per-tick divergence for each policy against the human, reported as a vector, never one score or ranking.逐 tick 计算每个策略相对人的偏离,以向量报告,绝不合成一个分数或排名。
- Knows the kinematics of the 14 bodies below.掌握下面 14 个本体的运动学。
Bodies we have models for有模型的本体
3D sheet →3D 一览 →Show the list展开列表
Loading the body index…
Rather talk now? 30 minutes想直接聊?30 分钟
RC Shadow · visitor page · demo buildRC Shadow · 访客页 · 演示版