Automation & Robotics
Refine
Year of publication
Document Type
- Conference Proceeding (51)
- Lecture (26)
- Article (21)
- Book (3)
- Part of a Book (2)
Language
- English (103) (remove)
Keywords
- Robotics (24)
- Materials (7)
- Bacteria (6)
- Robot (5)
- Saftey (5)
- Digital Factory (4)
- Object Detection (4)
- 3D (3)
- Agriculture (3)
- Digital Manufacturing (3)
Recent progress in machine learning and deep learning has enabled the implementation of plant and crop detection using systematic inspection of the leaf shapes and other morphological characters for identification systems for precision farming. However, the models used for this approach tend to become black-box models, in the sense that it is difficult to trace characters that are the base for the classification. The interpretability is therefore limited and the explanatory factors may not be based on reasonable visible characters. We investigate the explanatory factors of recent machine learning and deep learning models for plant classification tasks. Based on a Daucus carota and a Beta vulgaris image data set, we implement plant classification models and compare those models by their predictive performance as well as explainability. For comparison we implemented a feed forward convolutional neuronal network as a default model. To evaluate the performance, we trained an unsupervised Bayesian Gaussian process latent variable model as well as a convolutional autoencoder for feature extraction and rely on a support vector machine for classification. The explanatory factors of all models were extracted and analyzed. The experiments show, that feed forward convolutional neuronal networks (98.24%
and 96.10% mean accuracy) outperforms the Bayesian Gaussian process latent variable pipeline (92.08% and 94.31% mean accuracy) as well as the convolutional autoenceoder pipeline (92.38% and 93.28%
mean accuracy) based approaches in terms of classification accuracy, even though not significant for Beta vulgaris images. Additionally, we found that the neuronal network used biological uninterpretable image regions for the plant classification task. In contrast to that, the unsupervised learning models rely on explainable visual characters. We conclude that supervised convolutional neuronal networks must be used carefully to ensure biological interpretability. We recommend unsupervised machine learning, careful feature investigation, and statistical feature analysis for biological applications. View Full-Text
Keywords: deep learning; machine learning; plant leaf morphometrics; explainable AI
Abstract: The biological investigation of a population’s shape diversity using digital images is typi-
cally reliant on geometrical morphometrics, which is an approach based on user-defined landmarks.
In contrast to this traditional approach, the progress in deep learning has led to numerous applications
ranging from specimen identification to object detection. Typically, these models tend to become black
boxes, which limits the usage of recent deep learning models for biological applications. However, the
progress in explainable artificial intelligence tries to overcome this limitation. This study compares
the explanatory power of unsupervised machine learning models to traditional landmark-based
approaches for population structure investigation. We apply convolutional autoencoders as well
as Gaussian process latent variable models to two Nile tilapia datasets to investigate the latent
structure using consensus clustering. The explanatory factors of the machine learning models were
extracted and compared to generalized Procrustes analysis. Hypotheses based on the Bayes factor are
formulated to test the unambiguity of population diversity unveiled by the machine learning models.
The findings show that it is possible to obtain biologically meaningful results relying on unsupervised
machine learning. Furthermore we show that the machine learning models unveil latent structures
close to the true population clusters. We found that 80% of the true population clusters relying on
the convolutional autoencoder are significantly different to the remaining clusters. Similarly, 60% of
the true population clusters relying on the Gaussian process latent variable model are significantly
different. We conclude that the machine learning models outperform generalized Procrustes analysis,
where 16% of the population cluster was found to be significantly different. However, the applied
machine learning models still have limited biological explainability. We recommend further in-depth
investigations to unveil the explanatory factors in the used model.
Keywords: generalized procrustes analysis; machine learning; convolutional autoencoder; Gaussian
process latent variable models
Perspectives on Virtual Reality in Higher Education for Robotics and Related Engineering Disciplines
(2022)
Industrial engineering education has a strong focus on and affinity towards technology. While Virtual Reality hardware and applications advance and learning behaviour changes, it is particularly interesting to determine the possible use of Virtual Reality for teaching engineering subjects, for example fundamentals of robotics.
This paper presents a study which examines the possible use of Virtual Reality learning environments at higher learning institutions. The study shows perspectives of students and lecturers and identifies opportunities and challenges for the use of Virtual Reality in industrial engineering education. The results of the indicated study show that the participants have a positive attitude towards Virtual Reality and strong motivation for in class use. The study results also suggest, that Virtual Reality content creation should be included in engineering curricula.
This work represents the design and performance optimization of pumping aggregate for hydraulic active car
suspension systems. For solving of this task is required wide scope of interdisciplinary knowledge. The software used in
this project was SolidWorks from Dassault Systemes. Using this tool is possible to analyse and optimize the flow of
hydraulic fluid throw the electromotor of pumping aggregate. This papers shows among other, how to set the input
parameters and constraints such as pressure and velocity, how to simulate a rotating flow of cooling fluid inside
intermediate regions between stator and rotor. For approving a required lifetime of pumping aggregate a fatigue analysis
was done and represented above. The verification of simulation model and mandatory validation of simulation results
are made. The conclusions at the end of this work have confirmed the usage of computational fluid dynamic – software
for future researches of pumping aggregates.
Cyberphysical production systems are an important part of today’s manufacturing process. The ever-growing need of highly optimized,
i.e. at the same time flexible and ecient systems, requires the use of not only appropriate machines, but as well a
communication framework and data model that is manufacturer independent and scalable. This paper proposes a communicationframework
based on OPC UA that employs an agent-based architecture. The proposed system has been implemented and tested in
the Digital Factory of the UAS Technikum Wien. It shows promising behavior within distributed manufacturing systems.
In an effort to broaden the engineering applications of material extrusion based additive manufacturing (MEAM), new materials are being developed. Adding carbon-fibers (CF) has been one strategy to increase the mechanical performance of different thermoplastics. One challenge is to determine the amount of CF needed to increase the mechanical performance without affecting the “printability” of the compounds. In this paper, different amounts (10, 15, and 20 vol.%) of CF were added to recycled polypropylene (rPP) and polyamide 12 (PA12). A compatibilizer was used for rPP, but not for PA12. Filaments for MEAM were extruded from the different compounds and the viscosity as well as the tensile properties were measured and compared to the processed polymeric matrices. It was observed that the viscosities at the angular frequencies relevant for MEAM (100 to 200 rad/s) were not significantly different for rPP+CF compounds, but it was higher for PA12+CF compounds. As expected, the elongation at break significantly decreased with the addition of CF for all compounds. For the composites with an rPP matrix, the Young’s modulus and the ultimate tensile strength (UTS) continuously increased as the CF content increased to 20 vol.%. For PA12-based materials, the Young’s modulus and the UTS increased with CF content, but adding more than 15 vol.% did not further improve these values. Therefore, it was concluded that for PA12 the maximum amount of CF that should be added was 15 vol.%. Using scanning electron microscopy, it was observed that the CF were homogeneously dispersed in the rPP matrix, but not so well in the PA12 matrix, with fibers being more concentrated towards the rim of the filament. Finally, filaments of rPP, rPP+20CF, PA12 and PA12+15CF were used to print complex geometries by means of MEAM, and it was observed that CF helped to reduce the warpage compared to the unfilled filaments. A potential application of this phenomenon could be the reduction of the bed temperature to develop a more energy efficient MEAM process for semi-crystalline polymers.
Automatic Stereo Camera Calibration in Real-World Environments without Defined Calibration Objects
(2018)
A Novel and Automated Circle Pattern Recognition Technique for Infra-Red Stero Camera Calibration
(2013)
Digital Manufacturing
(2017)