Refine
Year of publication
- 2022 (3) (remove)
Document Type
- Master's Thesis (3) (remove)
Has Fulltext
- yes (3)
Is part of the Bibliography
- no (3)
Institute
- Fachbereich 1 - Institut Informatik (3) (remove)
Diese Arbeit beschäftigt sich mit der Erstellung einer Administratoroberfläche für die Lehre bei Photovoltaik (PV)-Praktika in der virtuellen Realität (VR). Die erstellte Umgebung bietet, mittels Bildschirmspiegelungen, Möglichkeiten zur didaktischen Anleitung und Unterstützung der Studierenden. Das Thema wurde aufgrund einer bestehenden Lehranwendung in der VR bedeutungsvoll und zeigt deutliches Potenzial. Diese Lehranwendung wird bereits umfassend und verpflichtend in den Praktika eingesetzt. Sie bietet einen praxisnahen Aufbau von Solaranlagen und erhöht gefahrlos die Experimentierfreudigkeit. Mit ihr lassen sich die aufgebauten Anlagen technisch prüfen, simulieren und bewerten. Zudem werden die beiden Möglichkeiten zur Unterstützung der Studierenden beurteilt. Als Ergebnis wird die Umsetzung der nahezu automatisierten Administratorober-fläche verdeutlicht und ein Usability-Test aus den Praktika evaluiert.
Schlagwörter: Administratoroberfläche, Bildschirmspiegelung, C, Didaktik, im-mersiv, Oculus Quest 2, Photovoltaik, Python, Tkinter, virtuelle Realität
Developing an intelligent chatbot that can imitate human-to-human interaction has become important in recent years. For this reason, many studies have been conducted to evaluate the quality of chatbots. Furthermore, various approaches and tools, such as sentiment analysis, have been created to improve the performance of chatbots.
This study examines previous research to identify the quality dimensions used to measure chatbots performance in order to develop a general chatbot assessment model that evaluates and compares chatbots quality. The developed evaluation model measures ten chatbot quality dimensions. This model is based on user experience. It requires human testers to interact with the chatbot to test its functioning and then a quantitative approach is used to collect data from user testing by conducting a survey with these testers. In this survey, they are instructed to evaluate the quality of the chatbot using a questionnaire that contains the items needed to evaluate each dimension.
This study also investigates whether sentiment analysis can improve the quality of chatbots and, if so, to identify the dimensions improved with sentiment analysis. For this reason, two chatbot versions are implemented using the Rasa framework (one that cannot detect sentiment and the other that analyzes sentiment and responds accordingly).
Following that, we used our evaluation model to evaluate and compare the two chatbot versions with two groups of participants by conducting a survey. In this survey, each group tested the functioning of one version. Then, both groups were instructed to use the items of the evaluation model to evaluate the version they tested. The goal of this survey was to evaluate the validity and reliability of the items used in the evaluation model to evaluate chatbots, and also to determine if sentiment analysis improved the chatbot quality by comparing survey results between the two groups. The results show that items used in the assessment model to evaluate chatbots are valid and reliable. The findings also indicate that sentiment analysis improves the chatbot’s quality. However, it improves the quality of some dimensions but not the majority of them.
Analyse von Unsicherheiten künstlicher neuronaler Netze und Integration in die Objektverfolgung
(2022)
Over the last few years, the development of assistance systems for motor vehicles has shifted from comfort functions to control tasks. Increasingly, these control tasks are also being transferred to semi-autonomous systems. One safety-critical aspect is the correct and reliable observation of the immediate environment by the vehicle. These observations can be used, among other things, to set up models for tracking objects. Due to recent research, topics such as uncertainties for object detections and the calibration of artificial neural networks are now emerging.
The goal of this work is to investigate the possibility of processing positional uncertainties of a detector in a multiple object tracking approach and the eects on the tracking of objects. Additionally, the calibration of the used detector will be evaluated and corrected if necessary. The eects of the calibration on the tracking results will also be investigated in this context. After an investigation of the procedure used to generate the position uncertainties of the detector, a connection to the multiple object tracking was made and an approach to process the uncertainties based on a Kalman filter was developed. The confidence of the detections was also remodeled. For this purpose, the confidence was interpreted as the existence probability and processed using a Bayes Filter to reflect the existence of the tracks. In addition, appropriate calibration methods for the position uncertainties and confidence were selected and incorporated into the tracking procedure. The validation of the presented approaches was performed on a data set for driving situations.
The evaluation of the results showed that a processing of the position uncertainties generated by a detector is feasible in the presented tracking approach. The interpretation of the confidence as existence probability leads to good results. Calibration of the confidence further improves the results. However, the calibration of the position uncertainties led to worse results. Further inves-tigation of other calibration methods for the position uncertainties is needed.
Keywords: Multiple Object Tracking, Kalman Filter, Neural Network Calibration