OmniContact: Contact-Rich Humanoid Object Interaction Motion Data

Motion capture data and processed humanoid trajectories for contact-rich box manipulation and soccer-style interactions. The dataset includes processed NPZ trajectories, original BVH motion, object poses, and contact labels. It was created by author lightcone02 and last updated on 2026-06-26.

Use Cases

• Training reinforcement learning policies for humanoid box manipulation based on motion capture trajectories.
• Simulating and analyzing contact dynamics in humanoid soccer-style interactions.
• Developing visualizers or tools for inspecting multi-modal motion data (BVH, NPZ, object pose).

Strengths

• Includes multiple synchronized data modalities: motion capture (BVH), processed trajectories (NPZ), object poses, and contact labels.
• Focuses on contact-rich interactions, a key challenge in humanoid robotics.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count and dataset scale are unknown, which may limit suitability assessment.
• Description metadata is limited; actual data quality requires manual inspection after download.

Provenance

Source

huggingface

Collection Method

Motion capture data collection and processing for humanoid trajectories.

Freshness

Last updated 2026-06-26 08:50:50; freshness should be verified.