Grounding Dino
Grounding DINO zero-shot object detection — natural-language
Sam3
Segment Anything 3 — text-, point-, and box-prompted instance
Curobo
NVIDIA cuRobo motion planning — collision-free trajectories to
Grasping With Planner
Top-down grasping via a fast axis-locked linear descend with a
Molmoact Libero
Run the MolmoAct LIBERO checkpoint
Gemini Er
Open-vocabulary 2D object detection via the Gemini Robotics-ER
Perceiving Object Parts
>
Tracking Objects
Long-running skill that drives the SAM3 tracker from the
Perceiving Objects Oneshot
>
Perceiving Next Item
>
Geometry
Pure-math 3D geometry toolbox — back-project masks and depth to
Pi05 Libero
Run the openpi π0.5 LIBERO checkpoint (pi05_libero) as a
Transporting Objects
Move the currently-held object to a destination and release. The
Vlm
Free-form and yes/no visual question answering against a hosted
Molmo
Visual pointing and Q&A via the Molmo VLM served from a
Grasping Short Axis
Deterministic short-axis-aligned grasp with CuRobo. The grasp
Grasping Direct Ik
Direct IK align-then-descend grasping. The gripper pre-rotates to
Perceiving Objects
>