DorianAtSchool/robocasa-trajectories-trajectory

    ---
    pretty_name: RoboCasa Trajectories Single
    configs:
    - config_name: default
      data_files:
      - split: train
        path: data/train-*
    ---

    # RoboCasa Trajectories Single

    This dataset contains one row per RoboCasa trajectory / episode.

    ## Structure

    Each row is one trajectory / episode.

    Episode-level JSON is stored inline:

• adapted_trajectory
• original_trajectory
• execution_metadata

Step-level data is stored in aligned sequence columns:

• step_index

• tool_name

• tool_args

• robot_idx

• success

    Image columns stay inline and viewable in the dataset table:
  
    - `room_view`
    - `top_view`
    - `map`
    - `agentview_center`
    - `agentview_left`
    - `agentview_right`
    - `wrist`
  
    ## Summary
  
    - rows: 48
    - episodes: 48
    - tasks: HotDogSetup, PrepareCoffee, PrepareSandwichStation
    - average steps per episode: 39.4
  
    ## Load
  
    ```python
    from datasets import load_dataset
  
    ds = load_dataset("DorianAtSchool/robocasa-trajectories-trajectory", split="train")
    ```
  
    ## Notes
  
    - Camera renders are stored as JPEG. Maps remain PNG.
    - MP4 videos are not included in the dataset.
    - Debug `initial` / `pre_initial_state` camera frames are not part of the dataset.
    - Row granularity: `trajectory`.
    - This layout is self-contained under `load_dataset()`, but nested sequence columns are less friendly for the HF table viewer.