Project “SGRP-AI”

Automated and Explainable AI for Tire Safety Inspection

The goal of SGRP-AI is to develop a industrial-scale, automated and explainable AI system for tire quality inspection using shearography. Shearography is a laser-based, non-destructive testing method that reveals even the smallest structural defects inside a tire.

Tire inspection is a safety-critical process: it is used both for spot-checking new tires and for the full inspection of every tire during retreading, ensuring that tires used in cars, trucks, and aircraft are free of hidden damage. Today, this evaluation is largely manual, time-consuming, and dependent on the experience of individual inspectors. SGRP-AI automates this step by combining deep learning, synthetic data generation (GANs), and explainable AI (XAI) into a system that delivers fast, objective, and traceable results — while making its decisions transparent and understandable through visual explanations.

By replacing subjective manual assessment with reproducible, AI-based evaluation, the project increases inspection throughput, addresses the growing shortage of skilled personnel, and builds trust through explainability. SGRP-AI is being developed in cooperation with SDS Systemtechnik GmbH.

Funding: Invest BW
06/2023 – 05/2026

Project team (alphabetical order):
Prof. Dr. Klaus Schlickenrieder. Focus: project lead, shearography, machine learning
Prof. Dr. Michael Munz. Focus: project co-lead, XAI, deep learning

Research assistants / doctoral students:
Manuel Friebolin. Focus: deep learning, dataset

News

  • New Project: AI-based tire testing
    We have a new cooperation project involving the company SDS Systemtechnik GmbH and the THU. Prof. Dr. Klaus Schlickenrieder, Faculty Production engineering and production management, is the project lead. News-Link: https://www.thu.de/de/Seiten/News_KI-Reifenpruefung.aspx

Press

https://www.thu.de/en/university/news/article/invest-bw-strengthens-research-cooperation-between-sds-systemtechnik-gmbh-and-ulm%20University%20of%20Technology

Related Publications

15935456 SGRP-AI 1 apa 50 date desc 1 1 title 371 https://aisd-ulm.de/wp-content/plugins/zotpress/
%7B%22status%22%3A%22success%22%2C%22updateneeded%22%3Afalse%2C%22instance%22%3Afalse%2C%22meta%22%3A%7B%22request_last%22%3A0%2C%22request_next%22%3A0%2C%22used_cache%22%3Atrue%7D%2C%22data%22%3A%5B%7B%22key%22%3A%228YKCX7UD%22%2C%22library%22%3A%7B%22id%22%3A15935456%7D%2C%22meta%22%3A%7B%22creatorSummary%22%3A%22Friebolin%20et%20al.%22%2C%22parsedDate%22%3A%222025-10%22%2C%22numChildren%22%3A1%7D%2C%22bib%22%3A%22%26lt%3Bdiv%20class%3D%26quot%3Bcsl-bib-body%26quot%3B%20style%3D%26quot%3Bline-height%3A%202%3B%20padding-left%3A%201em%3B%20text-indent%3A-1em%3B%26quot%3B%26gt%3B%5Cn%20%26lt%3Bdiv%20class%3D%26quot%3Bcsl-entry%26quot%3B%26gt%3BFriebolin%2C%20M.%2C%20Munz%2C%20M.%2C%20%26amp%3B%20Schlickenrieder%2C%20K.%20%282025%29.%20%26lt%3Ba%20class%3D%26%23039%3Bzp-ItemURL%26%23039%3B%20target%3D%26%23039%3B_blank%26%23039%3B%20href%3D%26%23039%3Bhttps%3A%5C%2F%5C%2Fwww.mdpi.com%5C%2F2673-2688%5C%2F6%5C%2F10%5C%2F275%26%23039%3B%26gt%3BGrad-CAM-Assisted%20Deep%20Learning%20for%20Mode%20Hop%20Localization%20in%20Shearographic%20Tire%20Inspection%26lt%3B%5C%2Fa%26gt%3B.%20%26lt%3Bi%26gt%3BAI%26lt%3B%5C%2Fi%26gt%3B%2C%20%26lt%3Bi%26gt%3B6%26lt%3B%5C%2Fi%26gt%3B%2810%29%2C%20275.%20https%3A%5C%2F%5C%2Fdoi.org%5C%2F10.3390%5C%2Fai6100275%20%26lt%3Ba%20title%3D%26%23039%3BCite%20in%20RIS%20Format%26%23039%3B%20class%3D%26%23039%3Bzp-CiteRIS%26%23039%3B%20data-zp-cite%3D%26%23039%3Bapi_user_id%3D15935456%26amp%3Bitem_key%3D8YKCX7UD%26%23039%3B%20href%3D%26%23039%3Bjavascript%3Avoid%280%29%3B%26%23039%3B%26gt%3BCite%26lt%3B%5C%2Fa%26gt%3B%20%26lt%3B%5C%2Fdiv%26gt%3B%5Cn%26lt%3B%5C%2Fdiv%26gt%3B%22%2C%22data%22%3A%7B%22itemType%22%3A%22journalArticle%22%2C%22title%22%3A%22Grad-CAM-Assisted%20Deep%20Learning%20for%20Mode%20Hop%20Localization%20in%20Shearographic%20Tire%20Inspection%22%2C%22creators%22%3A%5B%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Manuel%22%2C%22lastName%22%3A%22Friebolin%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Michael%22%2C%22lastName%22%3A%22Munz%22%7D%2C%7B%22creatorType%22%3A%22author%22%2C%22firstName%22%3A%22Klaus%22%2C%22lastName%22%3A%22Schlickenrieder%22%7D%5D%2C%22abstractNote%22%3A%22In%20shearography-based%20tire%20testing%2C%20so-called%20%5Cu201cMode%20Hops%5Cu201d%2C%20abrupt%20phase%20changes%20caused%20by%20laser%20mode%20changes%2C%20can%20lead%20to%20significant%20disturbances%20in%20the%20interference%20image%20analysis.%20These%20artifacts%20distort%20defect%20assessment%2C%20lead%20to%20retesting%20or%20false-positive%20decisions%2C%20and%2C%20thus%2C%20represent%20a%20significant%20hurdle%20for%20the%20automation%20of%20the%20shearography-based%20tire%20inspection%20process.%20This%20work%20proposes%20a%20deep%20learning%20workflow%20that%20combines%20a%20pretrained%2C%20optimized%20ResNet-50%20classifier%20with%20Grad-CAM%2C%20providing%20a%20practical%20and%20explainable%20solution%20for%20the%20reliable%20detection%20and%20localization%20of%20Mode%20Hops%20in%20shearographic%20tire%20inspection%20images.%20We%20trained%20the%20algorithm%20on%20an%20extensive%2C%20cross-machine%20dataset%20comprising%20more%20than%206.5%20million%20test%20images.%20The%20final%20deep%20learning%20model%20achieves%20a%20classification%20accuracy%20of%2099.67%25%2C%20a%20false-negative%20rate%20of%200.48%25%2C%20and%20a%20false-positive%20rate%20of%200.24%25.%20Applying%20a%20probability-based%20quadrant-repeat%20decision%20rule%20within%20the%20inspection%20process%20effectively%20reduces%20process-level%20false%20positives%20to%20zero%2C%20with%20an%20estimated%20probability%20of%20repetition%20of%20%5Cu22640.084%25.%20This%20statistically%20validated%20approach%20increases%20the%20overall%20inspection%20accuracy%20to%2099.83%25.%20The%20method%20allows%20the%20robust%20detection%20and%20localization%20of%20relevant%20Mode%20Hops%20and%20represents%20a%20significant%20contribution%20to%20explainable%2C%20AI-supported%20tire%20testing.%20It%20fulfills%20central%20requirements%20for%20the%20automation%20of%20shearography-based%20tire%20testing%20and%20contributes%20to%20the%20possible%20certification%20process%20of%20non-destructive%20testing%20methods%20in%20safety-critical%20industries.%22%2C%22date%22%3A%222025%5C%2F10%22%2C%22section%22%3A%22%22%2C%22partNumber%22%3A%22%22%2C%22partTitle%22%3A%22%22%2C%22DOI%22%3A%2210.3390%5C%2Fai6100275%22%2C%22citationKey%22%3A%22friebolinGradCAMAssistedDeepLearning2025%22%2C%22url%22%3A%22https%3A%5C%2F%5C%2Fwww.mdpi.com%5C%2F2673-2688%5C%2F6%5C%2F10%5C%2F275%22%2C%22PMID%22%3A%22%22%2C%22PMCID%22%3A%22%22%2C%22ISSN%22%3A%222673-2688%22%2C%22language%22%3A%22en%22%2C%22collections%22%3A%5B%5D%2C%22dateModified%22%3A%222026-09-21T06%3A06%3A05Z%22%7D%7D%5D%7D
Friebolin, M., Munz, M., & Schlickenrieder, K. (2025). Grad-CAM-Assisted Deep Learning for Mode Hop Localization in Shearographic Tire Inspection. AI, 6(10), 275. https://doi.org/10.3390/ai6100275 Cite
Scroll to Top