Parameters
Dict[str, Any]
Must include “gripper_name”; optional “num_grasps”, “planner” (“graspmoe”|“diffusion”), “grasp_threshold”, “camera_extrinsics”.
Union[str, np.ndarray, Sequence, None]
(N, 3) object point cloud.
Union[str, Image.Image, np.ndarray, None]
Depth image; together with seg_image and camera_intrinsics, the alternative input to point_cloud, converted to a point cloud server-side.
Union[str, Image.Image, np.ndarray, None]
Segmentation mask selecting the object; part of the depth-based alternative to point_cloud.
Union[str, np.ndarray, None]
3x3 camera intrinsics matrix; part of the depth-based alternative to point_cloud.
float | None
Optional HTTP timeout.
Returns
Dict[str, Any]: Dict with:- grasps: (N, 4, 4) array of grasp poses
- confidence: (N,) array of scores
- latency_ms: server-reported latency