MaNI-101: Real-World Robotic Manipulation Simulation Dataset

The MaNI series is a real-world simulation dataset for robotic manipulation, focusing on complex human-object interaction and fine-grained motion modeling. It systematically covers various manipulation types common in real-world robotic operation scenarios, including grasping, pushing, pulling, rotating, assembly, and tool use. The dataset models real-world contact, friction, constraints, and dynamics through high-fidelity physical simulation and was created by author bh2821.

Use Cases

• Training robotic manipulation policies based on simulated grasping and pushing actions.
• Modeling fine-grained motion and contact dynamics for tool-use scenarios.
• Benchmarking reinforcement learning algorithms on assembly and rotation tasks.
• Developing multi-modal perception models for human-object interaction understanding.

Strengths

• Covers multiple real-world manipulation types including grasping, pushing, pulling, rotating, assembly, and tool use.
• Models real-world physical properties like contact, friction, constraints, and dynamics through high-fidelity simulation.

Limitations

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

Provenance

Source

huggingface

Collection Method

High-fidelity physical simulation.

Freshness

Last updated 2026-02-10 02:41:38; freshness should be verified.