CalibratedDecisions.

Repo · Home, robots & devices

jev-libero

Fine-grained robot-manipulation study loop: TypeSafe Jev chooses layered intents/actions while local physics previews candidate effects on…

Open github.com ↗

From the repository

Fine-grained robot control with Jev, physics previews, and configurable LIBERO tasks.

emobodied-aijevliberollmmodel-based-controlmujocorobosuiterobot-manipulationrobotics

How builders describe it

Fine-grained robot control with Jev, physics previews, and configurable LIBERO tasks.
Fine-grained robot-manipulation study loop: TypeSafe Jev chooses layered intents/actions while local physics previews candidate effects on configurable LIBERO tasks.
A fine-grained robot control study loop for LIBERO tasks: Jev layers an intent, a contact or motion family, and one of 27 control inputs, while reversible MuJoCo previews evaluate each candidate before execution.

The decision Jev makes

Choose the device action or control input from the safe options.

Where it fits

Jev deciding things in the physical world: smart-home actions, robot and drone control, and devices that need an answer in milliseconds. All 31 home, robots & devices projects →

Related projects