Project

2025Public

STM Scan Quality Assessment Service

Project Brief

A multimodal perception service that evaluates STM scan quality from images and scan-parameter metadata.

Project Tech Stack

Core
multimodal scan-quality assessment
Interfaces
topography/current/lock-in images · scan-parameter metadata
Artifacts
quality judgments · diagnostic feedback · image-text pairs

Approach

The system pairs topography, current, and lock-in images with scan metadata to identify common cryogenic STM failure modes and produce compact diagnostic feedback.

Instead of pushing raw images and instrument detail upward, it compresses scan output into quality judgments and short explanations that higher-level agents can call as a perception service.

Within the autonomous-control stack, it serves as the perception layer between live instrument execution and high-level experimental planning.

Outcome

The service shortens debugging loops for delicate STM runs by turning multimodal scan data into callable quality assessments that downstream agents can act on.

Media

Multimodal STM Image–Metadata Model Training
Multimodal STM Image–Metadata Model Training