Machine Vision and Signal Processing

© Fraunhofer IPA

The Department of Machine Vision and Signal Processing develops and implements innovative system and application solutions for information processing in conjunction with technical processes. The focus of research and development work is on smart measurement and inspection systems (e.g., for zero-defect manufacturing) as well as high-performance automation and process optimization solutions. The department is specialized in the intelligent, automated interpretation of image and sensor information to solve complex tasks. Applications include among others:

  • specialized 2D measurement and inspection systems (e.g., with industrial image processing, thermography, ultrasound),
  • efficient 3D measurement and inspection technologies with industrial computed tomography and optical 3D sensors (e.g., laser line sensors, 3D smart cameras, time-of-flight cameras),
  • automated 3D object recognition and scene analysis for automated applications (e.g., bin-picking, belt-picking) and assembly, and
  • accurate monitoring of machine conditions and forecasting of quality parameters based on high-frequent machine signals (e.g., vibration, force, acoustics).

Key areas of activity

 

2D image processing

Zero-defect manufacturing is often only possible if each single produced product is inspected. Contactless image processing systems are ideally suited to handle such high-volume inspection tasks efficiently.

 

3D image processing

The use of 3D sensor technology in manufacturing is rapidly advancing thanks to ever faster and denser data acquisition. Our high-performance and intelligent 3D image processing methods result in innovative and highly accurate solutions in industrial measurement and inspection technologies, object recognition or scene analysis.

 

Non-destructive testing

Computed tomography (CT) can be used in quality assurance to generate a complete, non-destructive 3D image of a component with all its internal geometric features and structures. Consequently, material inspections and measurements are carried out simultaneously in one data set. Supplemented by intelligent image processing and thermographic measurement technology, we deliver comprehensive and highly-specialized inspection solutions.

 

Condition monitoring and quality forecasting

Condition monitoring and quality forecasting aim at detecting defects and faulty conditions in production at an early stage. The continuous analysis of sensor data and process information keeps production running at an optimum and prevents a decline in product quality. Our innovative methods can be universally applied in numerous branches of industry.

 

Motion analysis for optimizing processes

The analysis of the worker's movements, or scene analysis, makes it possible to improve processes in a human-centered way and to automate activities that do not add value. In addition, eye tracking can be used to capture the needs of the worker in data-driven user needs assessments. Typical areas of application include assembly assistance and training new employees, as well as automated documentation and touch-free control.

 

Image simulation and virtual measurement planning

Generating image data for AI-based analyses is a time-consuming and cost-intensive business. Images that are generated artificially but still appear realistic to sensors can considerably reduce the efforts and costs involved and significantly speed up the development of applications. Using simulated data, we also design industrial measurement and testing systems optimally tailored to the application in question.

 

Automatic optical inspection (AOI) with machine learning

The AI-based image processing system developed at Fraunhofer IPA extends existing optical quality controls. The resulting lower pseudo error rate in the AOI process means that fewer manual rechecks are required.

 

in German

AVaiL

The AVaiL joint research project aims to upgrade processes for the high-rate, cost-efficient and resource-saving production of aircraft structures made from metal. Transferring approaches already in successful use in the automotive sector aims at creating a flexible and cost-efficient automation environment. Due to the high diversity of components in the aerospace industry, system flexibility and the ability to network different systems are key aspects.

 

Flexible sheet metal test lab with TRUMPF

In a long-term cooperation, TRUMPF and Fraunhofer IPA have been working together to find out how artificial intelligence can be used to support the development of digital business models in sheet metal processing. The topics addressed include intralogistics and worker assistance as well as production control and data quality.

 

QU4LITY - Autonomous quality and zero-defect manufacturing

In the QU4LITY project, adaptable modules for detecting anomalies in process chains were developed based on the joint evaluation of data from different process steps. The modules serve as a basis for zero-defect manufacturing in complex process chains.

 

EU project QualiFibre

In the EU project QualiFibre led by Fraunhofer IPA, a software platform was developed for the joint analysis, display and fusion of data from the non-destructive methods of computed tomography, ultrasound and thermography. The focus was on inspecting carbon and glass fiber composite plastics.

 

in German

CFRP Extinguisher

Within a BMBF project, a lightweight and certifiable cabin fire extinguisher made of carbon fiber reinforced plastic (CFRP) was developed for use in air travel. The lower weight saves energy, thus also decreasing the amount of CO2 produced by air travel. Computed tomography and thermography were used as testing technologies for the development.

 

SynErgie

In the SynErgie project, AI-supported services are being developed so that production processes can be adapted to a volatile energy supply. The services automatically check whether production can be optimized at planning and machine level in order to take advantage of lower energy market prices. To do this, a variety of optimization and ML methods are used.

 

DeMoBat

The goal of the joint research project is to demonstrate that battery modules and e-drive units can be dismantled on an industrial scale using an automated process, while considering economic and regulatory conditions. The reliable recognition of objects and their position using optical sensor systems and intelligent recognition processes are essential if battery modules are to be successfully dismantled automatically.

 

Access Checker

A new measuring technique helps detect people infected with the coronavirus at a safe distance. It registers an elevated body temperature, raised pulse and rapid breathing without endangering the staff measuring these parameters.

Press release March 2021

New AI applications for medium-sized enterprises in Baden-Württemberg

Press release July 2020

Trumpf and Fraunhofer IPA ramping up artificial intelligence for industrial use

Press release June 2020

Employee-focused workplaces increase productivity

Press release April 2020 / in German

Access Checker

Press release June 2019 / in German

Thermal imaging cameras are gentle on sick animals

Press release April 2019 / in German

Artificial intelligence in quality assurance

“Machine learning and vision - a technological revolution thanks to artificial intelligence and modern image processing” - this is the motto of this year's Control event forum.

Annual report 2019, p. 21 / in German

Artificial intelligence in manufacturing

Article and overview of AI, its added value, benefits and potential for the manufacturing industry.

Project

SLEM

In the project “Self-learning and self-explanatory machine (SLEM)”, an intelligent assistance system is being developed that provides adaptive support for users with varying levels of knowledge and expertise.

Daten