RoboFAC: Multimodal VQA Dataset for Robotic Failure Analysis

RoboFAC is a multimodal visual question-answering dataset for robotic failure analysis and correction. It comprises over 10,000 robot manipulation videos and 78,623 question-answer pairs, supporting tasks across simulated and real-world environments. The dataset was created by MINT-SJTU.

Use Cases

• Train visual question-answering models based on robot manipulation videos.
• Benchmark failure reasoning algorithms for robotic systems.
• Develop multimodal AI agents capable of analyzing and correcting robotic failures.

Strengths

• Over 10,000 robot manipulation videos provide a substantial visual corpus.
• 78,623 question-answer pairs offer extensive annotations for VQA tasks.
• Covers both simulated and real-world environments, which may increase generalizability.

Limitations

• Column-level documentation is absent; field semantics must be inferred after download.
• Row count is unknown, which may limit suitability assessment.
• Last updated 2026-04-28 01:06:49; freshness should be verified.

Provenance

Source

MINT-SJTU

Freshness

2026-04-28 01:06:49