text stringlengths 0 11.8k |
|---|
1 |
[{'role': 'system', 'content': 'You are a helpful assistant that generates Python code to directly solve the task.'}, {'role': 'user', 'content': [{'type': 'text', 'text': '\nYou are controlling a Franka Emika robot with API described below.\nGoal: Lift the blue cube at least 10 cm and hold it there, then place the red... |
Model: qwen38-27b |
Config Path: /mmfs1/gscratch/zlab/rulins/robocasa-hyak/records/a19s/select/e4_tasks/configs/lift_then_stack_s4_m1.yaml |
Git Commit: 53e9966 (Dirty: True) |
Total number of trials: 100 |
Code generation success rate / Average reward / Task completed: |
0.970/0.323/17 |
Average code blocks: 8.190 |
Average regenerations: 7.190 |
Average finishes: 0.420 |
Elapsed time: 8004.12 seconds |
You are controlling a Franka Emika robot with API described below. |
Goal: Lift the blue cube at least 10 cm and hold it there, then place the red cube on top of the green cube. |
Success requires all of: the blue cube is raised at least 10 cm above its starting height at some point; at the end the blue cube rests on the table again and is not held; at the end the red cube rests on top of the green cube and is not held. |
You may write python code comments for reasoning but ONLY write the executable Python code and do not write it in code fences. |
After this code executes, you will receive the current console stdout and stderr (any errors or print statements made in the code). You will then have the chance to write new code to query the environment state and/or complete the task. |
The functions (APIs) below are already imported to the environment. If you want to use numpy, you need to import it explicitly. |
APIs: |
get_observation() -> dict[str, typing.Any] |
Doc: |
Get the observation of the environment. |
Returns: |
observation: |
A dictionary containing the observation of the environment. |
The dictionary contains the following keys: |
- ["robot0_robotview"]["images"]["rgb"]: Current color camera image as a numpy array of shape (H, W, 3), dtype uint8. |
- ["robot0_robotview"]["images"]["depth"]: Current depth camera image as a numpy array of shape (H, W), dtype float32. |
- ["robot0_robotview"]["intrinsics"]: Camera intrinsic matrix as a numpy array of shape (3, 3), dtype float64. |
- ["robot0_robotview"]["pose_mat"]: Camera extrinsic matrix as a numpy array of shape (4, 4), dtype float64. |
point_prompt_molmo(image: numpy.ndarray, text_prompt: str) -> dict[str, tuple[int | None, int | None]] |
Doc: |
Use Molmo to point to a coordinate in the image based on a text prompt. |
Args: |
image: np.ndarray: The RGB image to process. Shape: (H, W, 3), dtype uint8. |
text_prompt: str: The text prompt to point to. |
Returns: |
dict[str, tuple[int | None, int | None]]: Pixel coordinates for each |
object query; (None, None) if parsing failed. |
segment_sam3_text_prompt(rgb: numpy.ndarray, text_prompt: str) -> list[dict[str, typing.Any]] |
Doc: |
Run SAM3 segmentation on an RGB image conditioned on a text prompt. |
Args: |
rgb: |
RGB image array of shape (H, W, 3), dtype uint8. |
text_prompt: |
Text prompt for SAM3 segmentation. |
Returns: |
masks: |
A list of dictionaries. Each dict may contain: |
- "mask": np.ndarray of shape (H, W), dtype bool, |
where True means the pixel belongs to the instance. |
- "box": list [x1, y1, x2, y2] in pixel coordinates. |
- "score": float confidence score. |
segment_sam3_point_prompt(rgb: numpy.ndarray, point_coords: tuple[float, float]) -> list[dict[str, typing.Any]] |
Doc: |
Run SAM3 segmentation on an RGB image, optionally conditioned on an image coordinate point prompt. |
Args: |
rgb: |
RGB image array of shape (H, W, 3), dtype uint8. |
point_coords: |
(x, y) pixel coordinates of the point prompt. |
Returns: |
masks: |
A list of dictionaries. Each dict may contain: |
- "mask": np.ndarray of shape (H, W), dtype bool, |
where True means the pixel belongs to the instance. |
- "score": float confidence score. |
plan_grasp(depth: numpy.ndarray, intrinsics: numpy.ndarray, segmentation: numpy.ndarray) -> tuple[numpy.ndarray, numpy.ndarray] |
Doc: |
Plan grasp candidates using Contact-GraspNet for a single instance. |
This is a thin wrapper around the Contact-GraspNet planner. It does not |
apply any camera/world transforms or TCP offsets: the caller is |
responsible for transforming the resulting grasp poses into the desired |
frame and applying TCP offsets if necessary. |
Args: |
depth: |
Depth image in meters. |
Shape: (H, W), dtype float32/float64. |
intrinsics: |
Camera intrinsic matrix. |
Shape: (3, 3), dtype float64. |
segmentation: |
End of preview. Expand in Data Studio
No dataset card yet
- Downloads last month
- 66