Projects & Publications

Abstract

Breast reconstruction using autologous tissue, specifically with Deep Inferior Epigastric Artery Perforator (DIEP) flaps, has made significant advancements in recent years by aiming to minimize damage to the donor site. In this regard, selecting the appropriate perforator is crucial, given that the flap is perfused by a single perforator. The current standard method for selecting perforators is Computed Tomography Angiography (CTA), which has drawbacks including the use of contrast agents, exposure to radiation, high costs, and no information on flap perfusion. Recent studies have demonstrated that Dynamic Infrared Thermography (DIRT) is a non-invasive method capable of visualizing both the dominant perforators preoperatively and the perfused zones associated with these perforators intraoperatively. Identifying these perfused zones is essential for optimizing the breast's survival chances, but it requires an additional average surgical time of 60 minutes. The aim of this project is to predict perfused zones of specific perforators without intraoperative measurements, resulting in faster and more accurate treatments. This is achieved through the development of a Convolutional Neural Network trained with Finite Element Method (FEM) models of the abdomen with perforators. These models are constructed using both CTA and pre- and intraoperative DIRT data, with FEM updating to adapt the model's thermal behavior to the infrared measurements.

Researcher(s)

• Promoter: Steenackers Gunther
• Co-promoter: Thiessen Filip
• Fellow: Clarys Warre

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

Endoscopy plays a pivotal role in both diagnostic examinations and minimally invasive surgical procedures. A special type is wireless capsule endoscopy, where patients ingest a small pill-shaped camera. Despite its importance, the endoscopic images and videos exhibit serious drawbacks, such as substantial distortion, low resolution, missing frames, specular reflections, and so forth. In this project, I will tackle several of these challenges. In order to do so, I will first develop a novel endoscopic camera calibration procedure. Next, based on this, I will adopt approaches from the field of Gaussian process latent variable models and the world of generative AI in general to formulate models that construct an alternative latent space representation of the data. Probabilistic machine learning models, such as Gaussian processes, offer interpretability (no black-box), which is especially crucial in evidence-based medicine as it offers transparency and helps build trust with clinicians. Improved camera calibration and innovative perspectives on latent spaces hold the potential to revolutionise various techniques, including 3D trajectory estimation, mosaicking and many more. As a result, my research stands to significantly enhance the precision and efficiency of clinicians when interpreting endoscopic images. This, in turn, promises to elevate detection rates, enhance the accuracy of abnormality size measurements, and contribute to the advancement of minimally invasive surgery.

Researcher(s)

• Promoter: Penne Rudi
• Fellow: De Boi Ivan

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

In the worldwide R&D on computer-assisted and autonomous navigation the DDSHIP project will contribute by setting a new process flow methodology and test platform for validation and certification through investigations on: • more accurate and robust perception and situational awareness of the waterborne world around the ship in dense traffic and harsh weather conditions; • the accurate representation of the real behaviour of the ship in complex waterways with low under keel clearances and nearby banks and infrastructure; • the safe and smooth control of the ship through model predictive AI-trained controllers providing necessary collision avoidance. As accidents on waterways are mainly attributed to human actions in combination with failures of technical hard- and software or environmental circumstances, the support of captains, pilots or skippers on board the manned ship or the operator from a remote operation centre on an unmanned ship, this research should prove the capabilities of existing technologies (camera, sensors, manoeuvring model prediction, path-planning and steering) leading to smarter - more accurate and higher reliability – control.

Researcher(s)

• Promoter: Vanlanduit Steve
• Co-promoter: Hellinckx Peter

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

This proposal aims to improve and clinically validate our 3D-otoscopic device. Initially developed in POC project 38820, the device allows for real-time, in-vivo imaging of the eardrum surface. During preclinical trials at UZA, ENT physicians identified several issues that prevented accurate results. Firstly, physicians deemed color imaging necessary to navigate the device within the ear canal to quickly find the eardrum location. Secondly, the neural network responsible for converting the deformed fringe patterns to full-field height maps underperformed compared to lab settings because it was not trained on data representing the human eardrum's reflective properties. This project will resolve these issues so that the device can be used effectively in the clinical ENT setting. The goal is to elevate the device from TRL 4-5 to TRL 6 and prepare it for clinical validation and commercialization. This project leverages unique collaborations with ENT experts. It is protected by intellectual property, enabling us to publish our findings and establish the device's efficacy in clinical settings, thereby paving the way for future licensing opportunities. This phase is a critical step in positioning our technology as an essential tool for otoscopic examinations and middle ear pathology assessment.

Researcher(s)

• Promoter: Van der Jeught Sam
• Co-promoter: Dirckx Joris

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

Errors in radiotherapy can have significant consequences for patients and generate concerns in public opinion due to misconceptions surrounding ionizing radiation. To enhance its safety, the implementation of in vivo dosimetry is crucial. The VERIFIED project aims to advance individualized therapeutic procedures by utilizing patient-specific information, real-time dose, and deep learning techniques in adaptive radiotherapy (ART). The primary objective of the project is to develop dynamic end-to-end methods that closely simulate real patient treatments. Our project encompasses several key objectives. First, it involves the development and characterization of appropriate phantoms featuring movable and deformable inserts, specifically targeting lung and brain tumors for ART. Additionally, we focus on investigating individualized patient-specific real-time dosimetry in cases of non-small-cell lung cancer using Volumetric Modulated Arc Therapy (ART-VMAT). This approach enables accurate and timely monitoring of radiation doses. development of a realtime dose prediction protocol for non-small-cell lung and bladder tumors ART-VMAT. This protocol combines data obtained from the developed dynamic phantoms and the patient-specific real-time dosimetry system. Deep learning algorithms are employed to enhance the accuracy of dose prediction. Furthermore, an image-based system is being implemented to monitor the patient's head surface during in adaptive hypofractionated Gamma Knife radiosurgery (hfGKRS) for brain tumours, ensuring precise treatment delivery. Additionally, we will analyze the data obtained from the patient's head surface monitoring system, incorporating deep learning-based algorithms to generate a protocol for patient selection in hfGKRS.The proposed protocols integrate state-of-the-art deep learning methods with patient-specific real-time dosimetry in ART-VMAT and real-time position imaging in hfGKRS, effectively addressing several unmet needs in adaptive radiotherapy. These protocols encompass adaptability assessment, dosimetric verification, imaging validation, plan evaluation metrics, and treatment efficiency. By leveraging the power of real-time dosimetry, imaging, and deep learning, treatment efficacy can be enhanced while minimizing toxicity and radiation-induced side effects, ultimately resulting in improved patient outcomes in radiotherapy.

Researcher(s)

• Promoter: Vanlanduit Steve

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

The impending 31% surge in cancer incidences by 2030, coupled with a critical shortage of histopathologists, underscores an urgent need for innovations in diagnostic methodologies. A time-intensive aspect of histopathology is the staining of tissue slices, a pivotal step for disease diagnosis and research. Recently, hyperspectral imaging has been proposed to generate virtual stains on unstained tissues, a technique that could revolutionize tissue analysis. This method promises reduced errors, increased efficiency, multi-staining capabilities, and sample conservation. However, the technique is currently limited by small sample sizes, undefined wavelength band efficacy, and restricted data accessibility. This research project, aims to expand the sample size to 100 slices across four cancer types, employing three different hyperspectral cameras. We will create a comprehensive database, initially using the H&E stain as a reference. The project's second objective is to deploy deep learning algorithms to transform hyperspectral data into virtual stains and to ascertain the most effective wavelength bands. Finally, we aim to share our findings and dataset openly to encourage collaborative advancements. At InVilab, our infrastructure features an extensive array of imaging equipment, including a quantum cascade laser, enhancing our research capabilities in hyperspectral imaging. However, we currently face a shortfall in high-magnification lenses essential for detailed mid-to-long-wave infrared microscopy. An integral component for advancing our research. Securing funding of the BOF SRG will enable the acquisition of these critical lenses. This enhancement is imperative for integrating hyperspectral imaging into clinical practice, offering a strategic solution to the histopathologist shortage and advancing patient care outcomes.

Researcher(s)

• Promoter: De Kerf Thomas

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

Over the years there has been a tremendous evolution in breast reconstructions with free flaps, focusing on reducing donor site morbidity. Breast reconstructions with Deep Inferior Epigastric artery Perforator (DIEP) flaps have become the gold standard. As this flap is only perfused by a single perforator, the selection of the perforator is of main importance. Computed Tomography Arteriography (CTA) is the gold standard for the selection of perforators. However, this technique has some major drawbacks: the use of intravenous contrast, radiation, high costs, not being usable perioperatively, and no information on flow characteristics. Not only the selection of the perforators is mandatory for successful breast reconstructions with free flaps. Flap failure in breast reconstructions is often due to technical failures during the dissection of the perforator, failure of the anastomosis, or kinking or compression of the pedicle during flap-inset and shaping. Clinical monitoring is mostly used to diagnose these problems. During this study, we will further evaluate the use of Dynamic Infrared Thermography (DIRT) during breast reconstructions as an alternative non-invasive examination that is applicable during all phases of breast reconstruction. This technique allows for identifying the most dominant perforators and the area they perfuse. Our clinical study confirms that DIRT is capable to confirm the location of perforators of DIEP-flaps preoperatively. Moreover, using DIRT, extra information on the quality of the perforators is obtained by objective monitoring of flap perfusion with the same standardized measurement set-up. Our preliminary studies show that DIRT is a promising technique for selecting perforators and monitoring flap perfusion, used during all phases of breast reconstruction. The project goal is to further evaluate the use of DIRT during breast reconstructions in order to reduce flap failure and ultimately reduce the cost for our society.

Researcher(s)

• Promoter: Steenackers Gunther
• Co-promoter: Verhoeven Veronique

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

The need for protection and performance in various work and sport setups has driven the development of functional textiles and clothing with complex designs. Comfort is an important aspect for high performance sportswear and Personal Protective Equipment (PPE) as it affects the health, sports performance and work efficiency of athletes and workers. Unfortunately the comfort of these cloths, consisting of complex structures and special materials, is known to be poor. In many cases, the customers are satisfied with the functionality (i.e. protection against rain, cold, etc.) but disappointed in the poor comfort. Problem and innovation target: none of the existing state-of-the-art comfort test methods is ideal. They typically require numerous physical prototypes, specialized test equipment, lengthy and costly wearer tests, etc. Physiological parameters, such as skin temperature (Tsk) is linked to thermal sensations and comfort of the human subject and can be accurately assessed by Infrared (IR) thermography. Comfort is a complex matter influenced by the intensity of the activity, climatic conditions, textile material properties, clothing design and fit. Artificial Intelligence, in particular Deep Learning (DL), can be used to deal with these numerous parameters that influence the comfort. Goal and objectives: ComforTex-AI will help overcoming these issues related to comfort and aims to develop algorithms that will result in an affordable, quick and user-friendly methodology to assess and optimize the garment comfort. Fabric comfort characteristics (Ret/Rct), physiological parameters (i.e. Tsk), environmental factors (i.e. air temperature/humidity), work intensity and garment fit will be used as input for the AI Neural Networks (NN) that will predict comfort perceptions. The main expected result is a new algorithm-driven methodology developed in the form of a user-friendly tool for companies to aid in the design of comfortable functional clothing. Further results include among others: (1) Large library with quantitative data (i.e. Tsk, fabric properties Ret/Rct,) tested via state-of-the-art equipment and qualitative data (i.e. comfort sensations from human subjects) collected in various wear scenario's; (2) AI algorithm to assess and optimize clothing comfort (on 5-points scale); (3) Data-based product classification for specific usage conditions; (4) Validated cases for sportswear, workwear and PPE for specific purposes. Economic impact: implementation of the new methodology will assist the companies to make optimized choices of material and garment designs during the development phase. This will result in an increase of their turnover as a result of (1) shorter and cheaper development costs due to less physical prototypes and testing, (2) lower production costs due to more efficient use of materials and (3) merchandising of qualitative and comfortable clothing which comply with the wearer needs. This will furthermore limit the premature discharging, overproduction and overconsumption, which are currently huge concerns for the textile and clothing sector. Due to its strong multidisciplinary character, ComforTex-AI will enable the acquirement of new skills and knowledge and strengthen the market position of the companies in the sector. The specific target group consists of manufacturers of workwear, PPE and sportswear (20 to 25 companies in Belgium and 30 companies in Germany) and also producers of fabrics (11 weaving mills and 21 knitting mills in Belgium and 15 knitting mills in Germany). The project consortium consists of two sector associations (FKT as global coordinator and CREAMODA) and four RTOs with complementary expertise in materials and clothing comfort (HOGENT and Hohenstein Institute), IR thermography and AI techniques (UAntwerp and FITT / htw Saar).

Researcher(s)

• Promoter: Steenackers Gunther

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

Structured light profilometry is an established optical technique that measures the 3D shape of an object by projecting fringe patterns (usually lines) onto the object surface and by observing the deformed lines under a fixed angle. Today, state-of-the-art structured light profilometry requires three or more unique recordings to analytically determine the full-field height map of the object. This limits the 3D acquisition speed of the application, complicates the optical setup of the measurement system, and induces motion artifacts in the 3D scans when the target moves between subsequent recordings. In this project, we will train a custom neural network to convert a single deformed structured light pattern directly into its corresponding 3D surface map. By doing so, we will effectively solve the correspondence problem between deformed fringe pattern and 3D map using only a single input image. This will hugely increase the 3D measurement speed in real-time applications, with the frame rate of the camera now being the only limiting factor. In addition, the optical complexity and cost of current state-of-the-art optical scanning systems will be significantly reduced, which will create new possibilities in medical imaging, industrial inspection, machine vision, entertainment, and biometric access security applications. Furthermore, we will build on this new strategy to answer the question of whether neural networks can learn to extract high-resolution and absolute 3D information from a single 2D camera image of an object without using any projected lines, dots, or other fringe patterns – much like humans with monocular vision do. This will omit the need for a projection unit in a 3D scanner altogether and will effectively convert any smartphone camera, endoscope, or smart glasses into quantitative 3D scanning systems. This will result in an entirely new single-shot, ultrafast, and fully scalable 3D depth-sensing technique.

Researcher(s)

• Promoter: Van der Jeught Sam
• Fellow: Evans Rhys

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

In this project we propose a radically new plasma based-methodology to both characterize as well as treat cancerous tissue (with a focus on melanoma). We will use plasma excitation (in combination with laser vibration measurements) for in-situ characterization of the visco-elastic mechanical properties of biomedical tissue. These mechanical properties will allow us to detect and monitor cancerous tissue. Furthermore, we will develop a novel controlled plasma cancer treatment method which integrates the in-situ material identification method in order to tune the plasma therapy.

Researcher(s)

• Promoter: Vanlanduit Steve
• Co-promoter: Bogaerts Annemie
• Co-promoter: Dirckx Joris
• Co-promoter: Smits Evelien

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

Osteoarthritis (OA) of the knee is defined as a mechanically induced degenerative joint disorder. Lower-limb deformity is one of the main causes of disease development as load is one-sidedly shifted and unequally redistributed over the articular surface. Knee joint replacement surgery is frequently performed, however postoperative patient dissatisfaction is present in one fifth of patients. Therefore, the aim of this proposal is to deepen the understanding of joint-preserving procedures to ameliorate the native joint by addressing two important clinical obstacles: the static two-dimensional pre-operative planning and adverse effects on the adjacent ankle joint. Development of a discrete element analysis (DEA) model of the knee joint will be based on three-dimensional geometric morphometric data obtained from our in-house available, largest present multi statistic shape model (SSM). For dynamic assessment, data will be extrapolated out of parameterized gait analysis. The generated knee joint DEA model will be integrated with our DEA models of the hip and ankle to provide an entire lower-limb model. K-fold cross validation will be used for evaluation in terms of in-model accuracy, specificity and generalizability. Translation of the developed and validated lower-limb model from engineering science to the clinical field will enable preoperative planning to allow an optimal correction of the contact mechanics of the knee as well as the adjacent joints.

Researcher(s)

• Promoter: Steenackers Gunther

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

In this PhD research we will develop a non-destructive testing technique for the automated and quantitative inspection of bare and coated steel structures during manufacturing and operation, using a hyperspectral camera. Whereas the human eye and color cameras perceive three colors in the visible range, a hyperspectral camera is able to capture several tens of images over a wider wavelength range. This facilitates observation of phenomena that cannot be observed with traditional cameras. Although hyperspectral cameras have already proven their merit in quality control of food products, their use for non-destructive testing is still at its infancy. The main limitation of hyperspectral imaging is the limited spatial resolution. We will develop an image processing technique to artificially increase the spatial resolution of hyperspectral images. We will deploy a technique to continuously scan over the surface of the structure in order to reduce inspection times and we will apply the hyperspectral NDT technique to deblur images.

Researcher(s)

• Promoter: Vanlanduit Steve

Research team(s)

• Industrial Vision Lab (InViLab)

Project website

Project type(s)

• Research Project

Abstract

The goal of this Service Platform project is firstly to provide feasibility and data acquisition services to companies interested in utilising cutting-edge camera technologies (like high-speed, thermal and hyperspectral cameras). Our services will assist companies in making better, well-informed decisions and offer valuable perspectives on utilising and integrating these technologies into their solutions. Secondly, we will offer easy-to-use software tools needed to integrate cameras quickly and in a standardised way into their customised implementations. Our user-friendly generic camera acquisition toolbox for Python (GenPyCam) will facilitate camera deployment (also on embedded and virtualised systems). In this way, image processing companies will be able to solve problems they could not tackle before, their capabilities in terms of camera types will increase, and their camera software implementation cost will decrease significantly.

Researcher(s)

• Promoter: Vanlanduit Steve
• Co-promoter: Ribbens Bart

Research team(s)

• Industrial Vision Lab (InViLab)

Project website

Project type(s)

• Research Project

Abstract

In collaboration with a composer a conversion of wind power data to music was established. The data of 40 wind turbines in Antwerp was captured during a period of 1 year and used in the musical composition. The results were presented during a concert in Antwerp (Havenhuis).

Researcher(s)

• Promoter: Steenackers Gunther

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

The Port of Antwerp-Bruges is seeking solutions to perform accurate and fast detection of oil incidents in the port area. The port is polluted on a regular basis by oil which has a serious environmental and economical impact. Amongst the economical impacts are the costs for clean-up of the oil, which are related to its size, volume, the type of product and the location. These costs variate year to year, but in average we speak about the budget of 700K up to one million euro per year. The port is seeking solutions for early detection of the spills and follow up the spill while cleaning. Drones are seen as the ideal platform and are part of the port's future strategy towards digitization. To make drone inspections operational in 2021 is the priority of the Port of Antwerp-Bruges Authority. For this proposal, the port has identified two partners in Belgium, VITO and the University of Antwerp which have the necessary expertise to develop an appropriate solution. The objectives of this project are (1) to develop a prototype workflow to detect oil spills from a drone and (2) demonstrate the technology in the Port of Antwerp-Bruges. Port of Antwerp-Bruges Authority will launch a tender in Q3 2021 to select a company which will provide drone inspections in the port area. Automated oil spills detection is a part of the inspection program. Therefore it is very important for the Port of Antwerp-Bruges to help to develop this technology and to bring it to a higher TRL level. The Antwerp Port authority is co-funding by opening up their infrastructure, introducing an artificial spill and providing continuous feedback on the results. The result will be a prototype workflow to detect oil spills from drone images at TRL 5-6, accompanied by a protocol for the camera settings and flight protocol. The innovation in this proposal is linked to the challenging application of the technology in a complex and harsh port environment and the combined use of SWIR and RGB imagery.

Researcher(s)

• Promoter: Vanlanduit Steve
• Co-promoter: Scheunders Paul

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

In many fields of science and technology, the Midwave Infrared range (MWIR) (4000-400 cm-1) is a highly relevant part of the electromagnetic spectrum, because in this region many compounds and materials feature a unique pattern of absorption bands, directly related to their molecular structure. Current MWIR spectroscopic techniques (like FT-IR) are often used, but these are relatively slow and can only be used in laboratory conditions (on small samples). In this project we will use a Quantum Cascade Laser (QCL) to enable in-situ remote spectroscopy. Reflection (solids) or transmission (liquids and gasses) spectra can be measured with a detector by (fast) scanning of the QCL wavelength (up to 10.000 cm-1 per seconds). We will use the QCL laser in combination with three detectors available at the UAntwerp-InViLab and AXIS research groups: 1) a deuterated triglycine sulfate, or DTGS detector, 2) a scanning laser Doppler vibrometer to perform remote scanning photoacoustic spectroscopy, 3) a thermal camera for mid-wave hyperspectral imaging. The three proposed QCL-based systems are complementary to each other: the DTGS enables a very high wavelength resolution, the LDV can be used for photoacoustic spectroscopy to perform measurements at large stand-off distances (up to 100m), and the thermal camera-based setup delivers a very high spatial resolution, but with a lower wavelength resolution. The QCL based system will be used for the research at InViLab and AXIS in different applications: artwork, corrosion, biomedical and textile inspection. Furthermore, we have identified several other potential applications that we will look into in the future together with other UAntwerp research teams: plasma chemistry, histopathology, road materials, metal oxide powders, meso-porous materials, drug detection, recycling of polymer materials, wastewater.

Researcher(s)

• Promoter: Steenackers Gunther
• Co-promoter: Janssens Koen
• Co-promoter: Vanlanduit Steve

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

Oil spills in ports pose a direct risk to the environment and marine life, along with operational risks (sailing through the discharges, fouling of ships, fire risk) and high costs for the port itself. This project aims at an automatic detection, determination of severity and centralisation of communication based on advanced LWIR (longwave infrared) and multispectral SWIR (shortwave infrared) imaging via drones. Automation saves time, costs and creates a safer and cleaner port environment. and cleaner port environment.

Researcher(s)

• Promoter: Vanlanduit Steve
• Co-promoter: Scheunders Paul

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

In the last decades, metal additive manufacturing (AM) gradually evolved from a rapid prototyping technology to a promising transformative manufacturing technology. In the SBO-FWO DREAM project the UAntwerp research groups Visionlab and InViLab will tackle two important challenges that can transform AM into a reliable manufacturing solution for structurally loaded components: 1) the development of innovative online process monitoring solutions for increased part quality and 2) the development of advanced time-resolved computer tomography measurement techniques for part validation. The activities of the AM2 SEP research project will be integrated into the SBO-FWO DREAM project (if the re-submission of October 2021 is granted). A post-doctoral researcher will be hired to explore links between the technologies used in the two challenges of the project (process monitoring and part validation). In the case that the FWO SBO DREAM project is not granted, we will link with the AM group of VUB through the ongoing FWO SBO project HiPas, which has a similar scope, but with less focus on online process monitoring compared to the DREAM project. In addition, we will use the AM2 SEP budget to lift the SBO-FWO DREAM project to a European level (we have already identified two relevant calls: "HORIZON-CL4-2022-DIGITAL-EMERGING-01-05: AI, Data and Robotics for Industry optimisation", "HORIZON-CL4-2022-DIGITAL-EMERGING-01-03: Advanced multi-sensing systems"). To do so, we might hire a consultant to support us in the submission of a Horizon Europe RIA project. The work of the consultant will include performing a partner search, communication with partners and administrative support for the proposal preparation.

Researcher(s)

• Promoter: Vanlanduit Steve
• Co-promoter: De Beenhouwer Jan
• Co-promoter: Sijbers Jan

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

The overall objective of the project is to demonstrate the technical feasibility and economic added value of using of hyperspectral cameras for corrosion inspection and quality control of corrosion solutions (corrosion cleaning, wet and dry coatings and chemical surface treatments). The project focuses on the entire corrosion value chain. On the one hand, these are owners of infrastructure (petrochemical, port, energy), a group of some 100 mainly large Flemish companies, and on the other hand companies offering corrosion protection services (about 150). protection services (approximately 150 SMEs). For the rollout of the results, it is important that also companies that are active in corrosion inspection and camera integrators are also actively involved. These two groups of Flemish SMEs each count about 20 companies that, compared to the previously mentioned companies, have a larger R&D capacity to be able to implement the technology.

Researcher(s)

• Promoter: Vanlanduit Steve
• Co-promoter: Scheunders Paul

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

General conclusions and achievements The goal of Inspect 4.0 is to combine and integrate machine learning and machine vision technology into a flexible and accurate quality inspection system that can be deployed across a variety of textile manufacturing setups. · Use of existing and new vision technology for woven fabrics. Cameras were chosen and a setup was built to allow inspection with the vision technology on an inspection table similar to those already used in the industry. (KPI2: Woven and coated textiles from partners will be analyzed on the developed measuring setup) · Several cases were handled during the project. Our ambition was not to develop and use black-box models, but to build the models case-by-case and further optimize them with the test data. Based on the limited test data available, this has largely been achieved. · Use of existing and new vision technology for coated fabrics. Based on the initial test results, it was decided that measuring coated fabrics was not feasible with the current availability of cameras. This, combined with the fact that the user group was mainly interested in detecting production errors in woven fabrics, it was decided to no longer focus on this for the time being. · Development of a demonstrator. A woven fabric demonstrator was produced on a light table, in roll configuration. (KPI3: The demonstrators from the project will be widely known to the textile industry through workshops, seminars, professional literature and online videos). This will be used in the near future for workshops, dissemination days and fairs. · Carrying out a feasibility analysis. The technology was applied and demonstrated on-site for use in an industrial setting, with a partner selected from the guidance group. This was finally realized in collaboration with VDS Weaving. All practical research and development results were presented and documented in slides. Both the recordings of the online dissemination moments and the slides are available via https://www.centexbelpresents.be/en/inspect4-0. The code can be tested without obligation on request. The information was provided, provided with the necessary support from our researchers to familiarize the participating companies with the possibilities without focusing on the bells and whistles. Impact analysis The project has contributed to the development of knowledge on the use of camera vision technology for textile defect detection. It served as a platform to connect textile sector partners with technology companies. Despite the implementation of this knowledge at target group companies, more knowledge is still needed. Various events have provided space for interaction and future collaborations. The project attracted a lot of interest, with 48 companies and 6 non-profit organizations participating. There were a total of 125 individual contacts. The results were distributed through various channels, including LinkedIn and the project website. After each event, links to documents and surveys were sent by email to the target companies.

Researcher(s)

• Promoter: Steenackers Gunther

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

In this project, the research group InViLab develops machine learning techniques for the inspection of infrastructure (bridges, cranes, etc.) using images recorded by a drone. We will investigate how different camera types can be combined to improve the reliability.

Researcher(s)

• Promoter: Vanlanduit Steve

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

Precision-farming needs large-scale adoption to increase production at such a level that it significantly contributes to minimizing the gap between actual and required world-production of food. Increasing the measurement and actuation intervals of e.g. monitoring for pests and watering are expected to contribute to e.g. increased yields. Sensing is an important element to quantify productivity, product quality and to make decisions. Applications, such as mapping, surveillance, exploration and precision agriculture, require a reliable platform for remote sensing. In precision agriculture, the goal is to gather and analyze information about the variability of soil/water and plant conditions in order to maximize the efficiency of the farm field. This would also increase the burden on the farmer, as the measurement-time and data-processing time increases significantly. This can be mitigated with Automated (cooperative) Precision Farming with the use of autonomous driving vehicles, vessels, drones and dedicated installations mounted on regular agri-machinery. For the cooperative robotic missions, the data will be tagged with accurate position information and merged with other data in order to create a digital map. To achieve good performance for an intelligent system in autonomous navigation tasks we will also build a 3D world model which will be integrated with a digital twin at plant level in order to improve the local path such that we obtain accurate information. To integrate the data from heterogeneous sensors, a platform will be developed to determine the practicality of the available sensors for the optimization of the spatio-temporal interpolation. This project will focus on a single (standardized) platform where (robotic)paths, monitoring strategies can be set and the drones/USV's/AGV's automatically deployed when certain conditions are met. The measurement data will be available for different stakeholders in the same platform.

Researcher(s)

• Promoter: Vanlanduit Steve
• Co-promoter: Copot Cosmin

Research team(s)

• Industrial Vision Lab (InViLab)

Project website

Project type(s)

• Research Project

Abstract

The popular pinhole model for imaging sensors and the associated calibration procedures appear to be inadequate for some of the new generation sensor technology. Even for classical RGB cameras, this standard model leads to unstable calibrations, with the need for an extra model to remove lens distortion. We propose line varieties as a unifying modelling for a broad set of sensors. As opposed to other previously published attempts in this direction, we identify the sub-varieties that correspond to real sensors. This enables us to extend interpolation techniques and Gaussian processes, to support sensor calibration from small samples of lines. We aim fundamental contributions to the fields of Line Geometry and Probabilistic Numerics. Our goal is to develop the framework for multi-sensor configurations (laser scanners, IR-cameras,…), providing measurement fusion, using the developed line models, and to achieve accuracy levels for sensor-supported Radiotherapy.

Researcher(s)

• Promoter: Penne Rudi
• Fellow: De Boi Ivan

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

In this project, we will develop an easy to use method to monitor the thermal condition of a person as a function of time, with potential applications entailed in physical treatment or a sports activity. The method employs amongst others thermal imaging. To that end, we create a virtual 3D model of the person of interest. The proposed technique will enable the development of a flexible and mobile measurement system, which can be used in labs, hospitals, rehabilitation centers, sports training facilities, etc.

Researcher(s)

• Promoter: Vanlanduit Steve
• Co-promoter: Ribbens Bart
• Co-promoter: Sijbers Jan
• Co-promoter: Verwulgen Stijn

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

In the context of clinical treatment of cancers, a major challenge involves the precise delivery of therapeutic agents to the tumor while limiting off-target effects. This is true for multiple treatment modalities including radiotherapy and non-thermal plasma (NTP) therapy. Hence, the main focus of this research project is to introduce the design of a supervisory control structure into a patient-in-the-loop therapeutic application. This system will be developed by integrating 3 components: 1) environmental sensors, 2) a robotic control unit, and 3) a therapeutic device (NTP generator). Since NTP treatment is highly dependent on parameters such as treatment time, application distance, etc. a feedback approach is necessary to compensate for tumor motion induced by the patient during treatment (e.g. respiration). To this end, artificial intelligence tools, including neural networks, will be employed to model the dynamic disturbances of the tumor. The developed self-learning artificial intelligence models will be embedded within model-based controllers to predict and minimize the effect of disturbances. Performance of the control structure will be validated with real-time experiments of plasma delivery in biological systems. The proposed methodology has the potential to improve the precision and accuracy of clinical NTP treatment and consequently minimize damage to healthy tissue.

Researcher(s)

• Promoter: Copot Cosmin
• Co-promoter: Lin Abraham

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

The determination of local components in human skin from in-vivo measurements is crucial for medical applications, especially for aiding the diagnostic of skin diseases. In the study of skin cancer and burn wounds and more specifically as a methodology for diagnosis of cancer type and identification of skin penetration depth, it is of great relevance to investigate which cell types are present and how these are distributed at or below the skin surface. Consequently, a number of medical inspection techniques have been developed that can be used for the identification of malignant skin properties and more specifically skin cancer types. However, most of the existing techniques are increasingly contested because they either require destructive sampling (biopsy) or only measure on or under the skin surface (hyperspectral imaging) without identification of the penetration depth or detailed physiology of the maligned skin tissue. As a promising non-contact and non-destructive imaging technology, dynamic infrared thermography (DIRT) inspection will be used in combination with hyperspectral imaging (HSI) and physical modeling for fast and accurate skin property identification but also for assisted medical screening as it is possible to differentiate physiological properties based on a combined thermal-hyperspectral response of the skin. In order to optimize the accuracy and speed of tissue screening, the combined HS+IR measurement methodology will be assisted by numerical modeling.

Researcher(s)

• Promoter: Steenackers Gunther

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

In spite of its ability to successfully characterize the condition and materials of paintings and other works of art in a non-invasive way, Macro X-Ray Fluorescence imaging (MA-XRF) suffers from a drawback that significantly affects its most valued application: revealing hidden features and overpainted compositions. While the penetrative properties of the primary and secondary X-rays can be used beneficially to reveal subsurface information that is crucial for art historical scholars and conservators, the extent to which a particular layer can be visualized selectively depends on the exclusive presence of an element in that layer. By consequence, layers with a similar elemental signature emerge intermixed in the same distribution image while the exact layer sequence remains unclear. As a result, in many cases, (contested) sample extraction proves mandatory in order to assign the detected elements to a specific layer within the paint stratigraphy. In order to augment chemical imaging with an additional depth-dimension, a dual approach is presented: (1) separating surface signals from deeper signals by expanding the MA-XRF detector angle geometry and exploiting the resulting, potential depth information that lies within the absorption effects on emission line ratios, by adding a level of data-processing to the existing protocol; (2) reconstructing the layer buildup and allocation of the detected signals by including an Infrared thermographic camera (IRT). In order to characterize the number of layers present and their sequence, multi-sine heat excitation will be exploited for the spectral range of 1.5-5μm in combination with dedicated post-processing of the hypercube images in the frequency domain. The proposed multimodal MA-XRF+IRT measurement methodology is developed on paint mockups and validated on historical paintings and wood panels, in collaboration with the Royal Museum of Fine Arts Antwerp.

Researcher(s)

• Promoter: Steenackers Gunther
• Co-promoter: Van der Snickt Geert
• Fellow: Hillen Michaël

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

In this project we will use drones to detect and monitor damage in infrastructure: wind turbines, bridges, buildings, solar panels, pavements, etc. Firstly, an overview of available path planning tools will be given. Secondly, we will develop machine learning tools to automatically detect damage (cracks, potholes, corrosion). The third aim of the project is the development of a methodology to allow a systematic comparison of repeated drone based camera measurements. During the project 9 case studies will be performed. The project is performed by UAntwerpen and WTCB together with a large consortium of companies active in drone based inspections or owners of infrastructure.

Researcher(s)

• Promoter: Vanlanduit Steve
• Co-promoter: Van den bergh Wim
• Co-promoter: Vuye Cedric

Research team(s)

• Industrial Vision Lab (InViLab)

Project website

Project type(s)

• Research Project

Abstract

The necessity of designing flexible and versatile systems is one of the most current trends in robotic research. Including visual servoing techniques in an existing robotic system is a very challenging task. In this project a solution for extending the capabilities of a 6 DOF manipulator robot for visual servoing system development, is proposed. In order to achieve this task, different types of visual features (which can be extracted from the image using a visual sensor) are detected and their properties are analyzed. Here, visual features such as point features and image moments are taken into account for designing the controller. An image-based control architecture is designed and a real-time implementation on a manipulator robot is developed. The primary objective of this research project is to converge into an accurate algorithm for object reconstruction in a clutter environment and subsequently helping the robot to perform a visual servoing task. The object reconstruction is done by employing tools from artificial intelligence such as deep Convolutional Neural Network. The image acquisition and image processing together with the computing of the image-based control law will be implemented in Matlab. Thus, a new type of robot driving interface that links the robots' controller with Matlab environment is proposed. Such a user driver interface will allow not only to design and implement real-time controllers but also to perform other tasks such as identification, path planning, etc. Finally, the robustness and stability of the proposed visual feature based control law will be implemented, tested and validated in real-time through multiple experiments.

Researcher(s)

• Promoter: Copot Cosmin

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

Non-destructive testing using active thermography is still an expanding research area in order to achieve higher accuracy and faster measurements. More and more industrial manufacturers explore the opportunities of active thermography measurements resulting in more complex shapes and materials. Due to these evolutions it becomes nearly impossible to select the most applicable measurement setup in a fast manner. Especially inspections of large parts are a challenge since inspections of the complete part at once is not possible. Dynamic measurements are the solution to inspecting those samples, but consequently this implies new challenges regarding the measurement setup. In order to perform accurate inspections, trial and error is not a suitable solution because this working principle is time-consuming and should be redone every time the test sample changes, the measurement setup alters or when new innovations are discovered. The purpose of this research is to develop and implement an optimisation routine in order to give a suggestion of measurement setup parameters starting from finite element simulations and afterwards updating with knowledge of preliminary measurements. This optimisation routine will be performed using well-known response surface techniques and benchmarked with newly discovered methods. The optimisation routine will be tested on multiple samples in order to inspect the robustness and reliability.

Researcher(s)

• Promoter: Steenackers Gunther
• Fellow: Verspeek Simon

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

The consortium aims to achieve two main goals with this SBO project. First, through a rigorous research methodology to better understand how the roughness fatigue life of additive-made metallic components can be significantly improved. We anticipate this by rolling out a multidisciplinary approach, i.e. in terms of surface and shape metrology, non-invasive quantification of the residual stress and mapping of the process parameters that have an influence on the corrosion mechanisms. Moreover, a strong asset in this project is the possibility to also investigate the interrelationships between phenomena. The second main goal is to build up fundamental knowledge on how the laser-based hybrid production process can be substantially improved: In particular to be able to make complex-shaped metallic components with high precision and this without further intensive post-processing (in particular also known as " first-time-right approach).

Researcher(s)

• Promoter: Vanlanduit Steve

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

In this research project we want to take a further step in refining and concretising techniques for an automatic road surface quality inspection for application in the Flemish region, in order to simplify road management and save costs by having the road surface at the right times at the right times to renovate or renew places.

Researcher(s)

• Promoter: Vanlanduit Steve

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

Markerless motion tracking became very popular and common since the introduction of the Microsoft Kinect in 2010 in both the gaming community and industry. To use markerless motion tracking in the field of medical rehabilitation, a higher accuracy and reliability is needed. To achieve this goal, we will combine a 2-D skeleton detection algorithm with the data from multiple 3-D cameras. The developed procedure will be validated with the marker-based Vicon system of the M²OCEAN lab and calibrated 3D body scans of subjects in static position. Afterwards, the technique will be implemented on a treadmill to evaluate the gait of a person. To simulate real world gait information, subjects will wear virtual reality glasses. This virtual environment stimulates the brain and influences the gait of a person, which results in extra information compared to stand alone treadmill walking.

Researcher(s)

• Promoter: Penne Rudi
• Co-promoter: Ribbens Bart
• Co-promoter: Verwulgen Stijn

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

A Brushless DC Machine (BLDC) is the optimal motor to use in applications where a more or less constant, controlled, high rotational speed is required. Typical examples include driving: the compressor of a cooling system including refrigerators and air-conditioning, the propellers of a drone, fans and pumps in general, …. The BLDC is responsible for the lion share energy usage of these applications. Moreover, cooling systems consume a lot of energy worldwide because of their ubiquitous presence. On the other hand, for battery fed systems such as drones there is strong desire for increased autonomy. This means there is a strong desire to reduce the energy usage of BLDC driven systems. BLDC motors are typically driven with a square wave current. On the other hand, using sine wave currents could result in an energy efficiency increase of 10%. However, typical BLDC algorithms lack feedback to drive the machine with sine waves. Using an encoder to obtain this position feedback would increase the cost and complexity of the drive system and can be impossible due to limited mounting space. Therefore, so-called sensorless algorithms which estimate feedback signals based on easily measurable voltage and current signals, are of interest. Consequently, the central research question of this STIMPRO is formulated as: Develop and implement a sensorless algorithm to provide feedback for a BLDC drive algorithm using sinusoidal current waveforms and validate its energy saving potential. As a starting point this STIMPRO will consider an estimation algorithm, developed by the promotor, for stepping motors, to use in BLDC drives. This STIMPRO will be used as a kick-start to initiate electrical motor control research at UAntwerp. This project will serve as leverage to move the activities off the promotor in motor control, who started at ZAP at UAntwerp the 1st of September 2018, previously established at UGent to UAntwerp. To do so, the STIMPRO will be used to hire a researcher who will submit an FWO SB proposal. However, if FWO funding is rejected we will not finish this project empty handed. Given the work plan defined in the STIMPRO, and the experience of the promotor the project will certainly result in publications, a test bench, added experience for the hired researcher and the exploration of possible bilateral collaboration with Flemish companies on the subject. The work done in this STIMPRO will be beneficial for the Op3Mech research group as adding research on electrical motors is a vital in the broader robotics research. Moreover, the education on drivelines at the Faculty of Applied Engineering is currently not supported by academic research. Therefore, the research activities initiated in this STIMPRO are vital to continue education on these topics.

Researcher(s)

• Promoter: Derammelaere Stijn

Research team(s)

• Co-Design of Cyber-Physical Systems (Cosys-Lab)
• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

In this project a new methodology for product testing and quality control is developed based on infrared lock-in thermography. Infrared thermography permits to visualize the thermal/ warmup response of objects. In particular, lock-in thermography employs a sinusoidal light source to warm up the object being studied. Although pulsed thermography (PT) is commonly used as thermographic inspection technique, this method is not well suited for inspection of fragile structures (art and biological tissue inspection, blood circulation, …) due to the large instant energy emission which involves insufficient controllability and non-uniformity. On the other hand, with traditional lock-in thermography only one defect depth can be inspected at a time. In addition, at least one steady state period of the sine wave excitation is necessary to obtain quantitative results.

Researcher(s)

• Promoter: Steenackers Gunther
• Co-promoter: Janssens Koen
• Co-promoter: Louarroudi Ebrahim

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

Combining finite element models with non-destructive testing has enormous potential for valorization. The objective of this project is to develop a reliable damage detection and localization tool by combining NDT thermography data with FE modeling, making use of system identification. As the amount of experimental data is very high and depending on the resolution of the IR camera, the goal is to use virtual modeling in assistance of the NDT tests in order to gain accuracy and time-efficiency.

Researcher(s)

• Promoter: Steenackers Gunther
• Co-promoter: Lauriks Leen

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

In the project we use 3D vision techniques in order to make the assembly process more efficient and safer. This is done by tracking people, products and machines (like cobots) in a manufacturing environment.

Researcher(s)

• Promoter: Vanlanduit Steve
• Co-promoter: Penne Rudi

Research team(s)

• Industrial Vision Lab (InViLab)

Project website

Project type(s)

• Research Project

Abstract

Bone cancer affects children and young adults and requires wide removal of bone, leaving large defects. In order to save the limb and to restore its function in a lasting way, dead bone from bone banks or sterilised removed bone (graft) is used to fill the defect and is fixed by plates and screws. Still, in some patients a gap between the dead graft and the remaining living bone is seen, causing a delayed healing which leads to prolonged non-weight bearing periods (>1 year) and reoperations. We aim to reduce the healing time by introducing a predefined compression force to a graft, comparable to methods used in fracture fixation and megaprosthesis ingrowth. However, no literature is available evaluating the compression force and its effect on graft healing. Moreover, as bone cancer is extremely rare, this small patient group is often ignored for research funding to improve the current knowledge. We need to reproduce this compression force in a reliable way in different patients and different bone parts. Therefore we need to develop a standardised surgical procedure and determine the relation between the compression force and the surgical variables, eg screw positioning. Data from in vitro cyclic loading experiments and the patient's characteristics will be used for virtual simulation of compression force during level walking. These data will be essential for the future introduction and development of innovative techniques such as patient-specific instruments and implants.

Researcher(s)

• Promoter: Steenackers Gunther

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

The hip functions as a ball and socket joint, with cartilage layers that cover the joint surfaces on both sides protecting it from impacts and permitting smooth movements. When the cartilage is impaired by mechanical, infectious or inflammatory causes, the joint might eventually wear down - a disabling condition known as osteoarthritis. Recent literature indicates that up to 80% of all hip osteoarthritis cases might be related to subtle variations in the joint geometry.: These variations have been suggested to give rise to peak joint stresses and altered load distributions on the cartilage. Although the mechanism is getting increasingly recognized in the literature, profound understanding of its true impact is lacking. Further, the prevalence of these morphological variations is reported to be much higher than the actual number of patients presenting for treatment. The aim of this thesis is to explore the impact of variation in hip joint anatomy on load distribution during daily living activities. I intend to clarify the role of mechanical drivers in the onset and progression of cartilage lesions of the hip joint by means of advanced multidimensional statistics and personalized load and stress predictions. The final step of this thesis will be to gradually transfer these findings into clinical practice and at the operating theatre by providing virtual pre-surgical planning, accurately implemented during surgery, using state of the art navigation technology.

Researcher(s)

• Promoter: Steenackers Gunther
• Co-promoter: Audenaert Emmanuel

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

A new generation of digital cameras makes use of emitted light pulses, more precisely the time between the emission and the reception of the reflected pulse, for computing the depth of the viewed object. This "Time-of-Flight" principle is replacing other 3D-scan strategies such as stereovision and structured light. Though the concept and possibilities of a ToF-camera essentially differs from these that are offered by "classical" optical cameras, the computer vision community still falls back on proven methods for calibration and structure-from-motion issues. We propose new techniques, fully exploiting the Time-of-Flight power, avoiding detection and recognition of features in the image. In a further step, we intend to design a new camera model, more general than the familiar pinhole model, providing a uniform framework for both lateral as depth calibration of ToF-cameras. The theory will be validated by simulations and real experiments (executed by a computer driven robot manipulator). Finally, real life applications will be considered, in cooperation with some of our industrial partners.

Researcher(s)

• Promoter: Penne Rudi
• Co-promoter: Steenackers Gunther
• Fellow: Roios Pedro

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

In the study of historical paintings and more specifically as a preparation for restoration activities of such artefacts, it is of great relevance to investigate which materials and degradation products are present and how these are distributed at or below the painting surface. Commonly used non-destructive in situ methods such as X-ray fluorescence (XRF) and X-ray diffraction (XRD), are only used for spot analyses and require several minutes to record a spectrum from a single sample position, resulting in long scanning times required to record the data hypercubes. As an alternative, thermography inspection, as a non-contact and non-destructive technique is used for material parameter identification but also for art inspection as it is possible to differentiate chemical compounds. Therefore the goal of this research proposal is to improve non-invasive macroscopic material characterization of flat objects, both from an industrial and cultural heritage context, by augmenting existing elemental imaging technology with more species specific imaging of organic and inorganic compounds and this by combining the established X-ray based approaches with IR thermography and hyperspectral (HS) images. A combined X-ray, IR thermography and HS technique eliminates the disadvantages of these techniques and results in a faster measurement and material identification technique with respect to measurement time but also accuracy of the material parameter identification.

Researcher(s)

• Promoter: Steenackers Gunther
• Co-promoter: Janssens Koen
• Co-promoter: Ribbens Bart

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

Rotating machines appear in many application fields ranging from large scale applications (e.g. wind turbines) to smaller ones (e.g. medical fluid pumps). The availability of a mathematical model for the dynamical behavior is of crucial importance for the design, prediction and control of these rotating systems. In the scientific domain of "system identification", the mathematical model of the system under test is retrieved through experimental input-output data. Since the dynamical characterization of rotating machines is non-linear as well as time-varying, it cannot be modeled adequately using classical existing estimation or identification methods. The aim of this project is then to develop a theoretical framework to model the time-varying and non-linear dynamical behavior of rotating machinery from experimental data. The proposed methodology consists of modeling the non-linear and time-varying dynamical character of rotating devices through a collection of linear periodically time-varying models. In this project, we will focus on the identification and validation of the non-linear and time-varying dynamics of a mechanical rotor suspended on hydrodynamic plain bearings. The novel approach consists of four main steps: (i) Construction of a non-linear and time-varying virtual model of "fluid-driven" bearing—rotor systems starting from the laws of physics; (ii) Development of a parametric identification technique; (iii) realization and adjustments of the controllable rotor—bearing setup; (iv) validation of the theoretical framework on the real-life rotor—bearing setup.

Researcher(s)

• Promoter: Louarroudi Ebrahim
• Promoter: Vanlanduit Steve
• Co-promoter: Copot Cosmin
• Co-promoter: Vanlanduit Steve

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

Camera positioning in vision applications is challenging, but of crucial importance. This is definitely the case when the goal is to make a 3D scan. This is because camera positions determine which parts of an object are visible and which measuring accuracy will be achieved. Our ambition is to automatically determine the scan path or positions of the camera during a 3D scan for a known object. To solve this engineering problem we will use mathematical techniques as 'tensor voting' and 'surface fitting'. The final algorithm provides the industry with the following advantages: 1. Faster and more efficient 3D scans 2. More complicated objects can be scanned 3. Automatic scan planning for every type of 3D sensor/camera in one model. 4. Automatic scan planning for specific measurement setups currently used in industry. This reduces the need for expensive experts.

Researcher(s)

• Promoter: Penne Rudi
• Fellow: Bogaerts Boris

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

In this project we will develop a technique that combines information from 3D time-of-flight camera's and computer-aided-design drawings of a product in order to facilitate product testing and quality control. The proposed procedure firstly allows the test engineer to automatically determine the sensor positions on a product. Secondly, we develop a methodology to perform vibration measurements on moving components (wind turbines, wind screen wipers, etc. or products on a conveyor belt).

Researcher(s)

• Promoter: Vanlanduit Steve
• Co-promoter: Mertens Luc

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

Quasi time-periodic phenomena show up in many engineering fields. One could think of wind turbines or helicopters with rotational speed fluctuations, the vibrations and acoustic noise generated in combustion engines, the electrical impedance of a living heart with heart rate variability for cardio-vascular monitoring, respiratory systems with breathing rate variability, to name a few. Those systems in engineering have the special property that their dynamic behavior changes quasi-periodically over time. The irregularity of the periodicity in the above-mentioned applications can be faithfully modeled by virtue of a periodically time-varying model with varying periodicity. This way of modeling bridges the gap between the well established identification framework for linear time-invariant systems and the more complex approaches for non-linear time-variant systems. The extraction of experimental quasi time-periodic models in the frequency domain meant for physical interpretation, analysis, prediction or control can be a useful step for the practicing engineer. Hence, the main focus of this project will be the development of a generalized identification framework for quasi time-periodic systems with applications in the mechanical and bio-medical engineering.

Researcher(s)

• Promoter: Vanlanduit Steve
• Fellow: Louarroudi Ebrahim

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

A new generation of digital cameras makes use of emitted light pulses, more precisely the time between the emission and the reception of the reflected pulse, for computing the depth of the viewed object. This "Time-of-Flight" principle is replacing other 3D-scan strategies such as stereovision and structured light. Though the concept and possibilities of a ToF-camera essentially differs from these that are offered by "classical" optical cameras, the computer vision community still falls back on proven methods for calibration and structure-from-motion issues. We propose new techniques, fully exploiting the Time-of-Flight power, avoiding detection and recognition of features in the image. In a further step, we intend to design a new camera model, more general than the familiar pinhole model, providing a uniform framework for both lateral as depth calibration of ToF-cameras. The theory will be validated by simulations and real experiments (executed by a computer driven robot manipulator). Finally, real life applications will be considered, in cooperation with some of our industrial partners.

Researcher(s)

• Promoter: Penne Rudi
• Co-promoter: Mertens Luc
• Fellow: Bogaerts Boris

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

Due to a lack of standardization, it is impossible for Time-of-Flight (ToF) camera users to come to an objective evaluation during a benchmarking. This project aims to create an objective method for the comparison of Time-of-Flight (ToF) cameras. This can be done by performing controlled measurements using cameras mounted on a robot arm to study the influence of the integration time, reflection coefficients and the incident angle.

Researcher(s)

• Promoter: Van Barel Gregory

Research team(s)

• Industrial Vision Lab (InViLab)

Project website

Project type(s)

• Research Project

Abstract

The overall goal is to develop an automated process that is able to increase the intake rate of empty-bottles. At this moment an operator is full occupied by this task and maybe this could be changed in a system in which the operator is nearby but become able to do other tasks meanwhile. The correct classification of Ordal bottles with respect to all foreign bottles forms a challenge and would end in an increase in production. Suspicious containers are handled through a bypass.

Researcher(s)

• Promoter: Mertens Luc

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

This project represents a research agreement between the UA and on the onther hand IWT. UA provides IWT research results mentioned in the title of the project under the conditions as stipulated in this contract.

Researcher(s)

• Promoter: Mertens Luc
• Co-promoter: Penne Rudi
• Co-promoter: Steenackers Gunther

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

The hip functions as a ball and socket joint, with cartilage layers that cover the joint surfaces on both sides protecting it from impacts and permitting smooth movements. When the cartilage is impaired by mechanical, infectious or inflammatory causes, the joint might eventually wear down - a disabling condition known as osteoarthritis. Recent literature indicates that up to 80% of all hip osteoarthritis cases might be related to subtle variations in the joint geometry. These variations have been suggested to give rise to peak joint stresses and altered load distributions on the cartilage. Although the mechanism is getting increasingly recognized in the literature, profound understanding of its true impact is lacking. Further, the prevalence of these morphological variations is reported to be much higher than the actual number of patients presenting for treatment. The aim of this thesis is to explore the impact of variation in hip joint anatomy on load distribution during daily living activities. I intend to clarify the role of mechanical drivers in the onset and progression of cartilage lesions of the hip joint by means of advanced multidimensional statistics and personalized load and stress predictions. The final step of this thesis will be to gradually transfer these findings into clinical practice and at the operating theatre by providing virtual pre-surgical planning, accurately implemented during surgery, using state of the art navigation technology.

Researcher(s)

• Promoter: Steenackers Gunther
• Co-promoter: Audenaert Emmanuel

Research team(s)

• Industrial Vision Lab (InViLab)

Project type(s)

• Research Project

Abstract

We focus on 2D point sets with an elliptic shape and 3D point sets with an ellipsoidal shape, e.g. in camera images or a data fusion setting. Noise on these data points forces us to look for robust procedures that derive the quantities we need. Motivating case study: Suppose that the image is taken by a calibrated camera from a ball with known radius, what is the position of this ball relative to the camera?

Researcher(s)

• Promoter: Penne Rudi
• Co-promoter: Mertens Luc
• Co-promoter: Steenackers Gunther

Research team(s)

• Industrial Vision Lab (InViLab)

Project website

Project type(s)

• Research Project