ViFailback: Robotic Manipulation Failures with Visual Symbol Guidance

ViFailback is a real-world robotic manipulation failure dataset introduced by sii-rhos-ai in the 2025/2026 research paper 'Diagnose, Correct, and Learn from Manipulation Failures via Visual Symbols'. It provides image-based data focused on diagnosing and correcting errors in robotic tasks through a specialized visual symbol interface. The dataset is associated with Arxiv paper 2512.02787.

Use Cases

• Training diagnostic models to identify robotic failures using visual symbols
• Developing robotic error correction policies based on visual feedback
• Evaluating visual symbol interfaces for human-in-the-loop robotic learning

Strengths

• Real-world robotic manipulation data rather than synthetic simulations
• Includes visual symbol guidance for failure diagnosis
• Directly linked to Arxiv research paper 2512.02787 for methodological context

Limitations

• Unknown total record count and file size
• Lack of detailed column descriptions in the metadata
• Geographic bias toward the United States as indicated by region tags

Provenance

Source

sii-rhos-ai (Arxiv 2512.02787)

Collection Method

Real-world robotic sensors and visual symbol annotation

Freshness

Last updated March 2026.

Geography

United States