Keynote Talks
Wednesday, Sept. 20, 14:30-15:30, Plenary Hall
Leap into the Industrial Metaverse: Where Wearables and Sensors Meet Reality
Dr. Hendrik Witt, TeamViewer
|
Abstract: Over the past few years, we saw a massive boost of digitalization in white-collar jobs. However, 80% of today’s workforce are desk-less workers, working for example in logistics, manufacturing, or field service. To reach the next level of digitalization and tap into this vast potential, the question is, how do we empower desk-less workers? And why is the Metaverse a huge opportunity in digitalizing the frontline workforce? TeamViewer strongly believes in a continuous convergence between the real world, the Internet and technology in general. This is how the Metaverse is vastly defined. And this is also how we designed our AR platform Frontline which is running on smart glasses as well as smartphones, tablets and other wearables and that offers a comprehensive set of capabilities around Augmented-, Mixed Reality and Artificial Intelligence. In his keynote, Hendrik Witt, Chief Product Officer at TeamViewer, will share real-world use cases and address how companies are using AR solutions to optimize processes. Get an inside look at the technology driving the next industrial revolution and learn why wearables and sensors are paramount to the success of the Industrial Metaverse. |
Biography
Dr. Hendrik Witt
Thursday, Sept. 21, 9:00-10:00, Plenary Hall
Alicia & Bernhard: A Communication Love-story Through the Ages
Prof. Dr. Nilesh Madhu, IDLab, Department of Electronics and Information Systems, Ghent University
|
Abstract: Communication is the lynchpin of civilisation, and effective communication needs to be clear, unambiguous, and timely. The quest to achieve this is as old as the hills and will continue to occupy the interest of many generations of engineers to come. In this talk we shall follow our intrepid, chatty pair Alicia and Bernhard as they encounter the typical hurdles to good communication and see how our community has worked unceasingly to help them overcome these obstacles over the years. The past two decades alone have seen our approaches evolve from using painstakingly hand-crafted, stochastic models to the plethora of impressive data-driven approaches, that is the state-of-the-art today - a truly spectacular leap, with equally spectacular results. However, can we consider all challenges and obstacles now overcome? Thankfully, as we shall see, no! Just as we peel back one layer of the communications "onion", another layer presents itself - offering new possibilities and opportunities to push the boundaries of the field inexorably forward. While gracefully acknowledging the gratitude of Alicia and Bernhard, and recognising the exciting path that lies ahead, I'd also like us to reflect on whether the knowledge and experience of the past bring us anything in facing these challenges. Or are the new data-driven paradigms sufficient on their own? |
Biography
Prof. Dr. Nilesh Madhu
Friday, Sept. 22, 9:00-10:00, Plenary Hall
Information Extraction from Speech and Audio Sensor Data - Challenges for Applied Research in the Security Domain
Prof. Dr. Frank Kurth, Fraunhofer FKIE, Dept. Communication Systems
|
Abstract: Information extraction from sensor signals plays a crucial role in the domain of security-critical applications. As documented not only in current international conflicts and in the area of organized crime, speech communication still plays a major role in this domain. On the side of security authorities, this requires a reliable and complete information extraction of audio signals acquired using various sensor sources in compliance with the respective laws and regulations. This kind of information extraction requires the interplay of a whole repertory of audio and speech processing methods ranging from preprocessing methods, source separation, and voice activity detection to automatic speech recognition and machine translation. A particular challenge in such scenarios is the usually poor quality of the audio signals recorded in "non-cooperative" settings. Furthermore, such monitoring scenarios usually produce big data volumes to be processed whereas the application might require near real-time processing facilitating efficient decision support. This talk presents several concrete application scenarios from the security domain requiring information extraction from speech and audio signals. Different approaches based on both classical signal processing and modern deep learning techniques are proposed. The focus of this talk is to highlight, as "lessons learned", some practical project experiences of the last years, thus putting into perspective results from basic research and requirements of real-world application scenarios. |
Biography
Prof. Dr. Frank Kurth
