Sign in to view source links and access this dataset
Description
Isaac Sim observation images for frame-level high-level action selection in robotic task planning. Each sample includes an RGB observation image, a task instruction, a frame-level high-level action label, an action type, and an optional target object. The dataset was created by Kuoskyler and was last updated on 2026-05-21.
Use Cases
Training models to select the next high-level action based on the current observation and task instruction.
Benchmarking visual task planning algorithms using simulated RGB observation images.
Studying the relationship between visual scenes, textual instructions, and high-level action labels.
Strengths
Dataset is designed for a specific research problem: execution-time high-level decision making in robotics.
Each sample contains multiple aligned data types: an RGB image, a text instruction, and structured action labels.
Limitations
Column-level documentation is absent; field semantics must be inferred after download.
Row count is unknown, which may limit suitability assessment.
Last updated 2026-05-21 09:50:37; freshness should be verified.
Provenance
Source
Kuoskyler on Hugging Face.
Collection Method
Likely generated using the NVIDIA Isaac Sim robotics simulation environment.
Freshness
Last updated 2026-05-21 09:50:37.
License is unknown; terms of use must be verified before application.