Prof. Sule Yildirim Yayilgan
Professor Sule Yildirim Yayilgan works at the Department of Information Security and Communication Technology (IIK), NTNU since 2009. She received a MSc. degree in Computer Engineering in 1995, and PhD in Artificial Intelligence and Computer Science in 2002. She has worked more than 25 years in academia teaching and doing research and served as Head of the Department between 2005-2009. She has participated in projects funded by EU Horizon 2020, Eurostars, Erasmus+ programs, the Research Council of Norway, the Regional Research Council of Norway and the Ministry of Foreign Affairs, Norway. She belongs to the Center fo Cyber Information Security (ccis.no) and she is leading the research group MR PET: Multidisciplinary Research group on Privacy and data protEcTion. She has been supervising students at different academic levels and has been publishing more than 100 journal and conference papers.
Speaker: Prof. Sule Yildirim Yayilgan
Affiliations: Norwegian University of Science and Technology, Norway
Speaker: Prof. Viet-Hung Vu
Affiliations: Royal Military College of Canada
Prof. Viet-Hung Vu
Canadian professional engineer and PhD in 2010 with the prestigious Canada NSERC Innovation Challenge Award. Currently, Dr. Vu is an Assistant professor at the Royal Military College of Canada and an adjunct professor at École de technologie supérieure. With the NSERC Discovery grant, he and his research team pursue innovations in robot structures’ dynamics and operational modal analysis. The approaches include both analytical modeling and experimental analysis and a hybrid of the two. Some recent findings are the linearization and symbolic implementation of dynamic equations, the times series model updating for automatic operational modal analysis, the forces identification, and the uncertainties in operational modal analysis.
Abstract: Recent studies and research on the dynamics and vibration of an industry-developed flexible robot with 6 degrees of freedom are presented. The flexibility of the joints and links is considered. The nonlinear dynamics and non-stationary vibration behavior of the robot structure during a grinding application are studied in both analytical and experimental approaches. A linearization model is presented, in conjunction with a symbolic computation, allowing the direct and inverse analytical dynamics of the system in the whole workspace. In the experiments, modal testing and operational modal analysis using the autoregressive moving average models and various identification techniques reveal the true spatial dynamical properties and the updating of these parameters in non-stationary vibrations. The two approaches’ efficient performance and the matching of the analytical-experimental results promise a perspective of a hybrid modal analysis technique for flexible structures and robots under real, complex working conditions.
Prof. Evangelos Pournaras
Dr. Evangelos Pournaras is Associate Professor in the School of Computing at University of Leeds, UK, where he leads the Distributed Intelligent Social Computing (DISC) lab. He is a also UKRI Future Leaders Fellow, an Alan Turing Fellow and a research associate at UCL Center of Blockchain Technologies. Evangelos' research interests focus on distributed and intelligent social computing systems with expertise in socio-technical domains of Smart Cities. He has more than 5 years of research experience at ETH Zurich after having completed his PhD studies at Delft University of Technology. Evangelos has also been a visiting researcher at EPFL and has industry experience at IBM T.J. Watson Research Center. Evangelos has won the Augmented Democracy Prize, the 1st prize at ETH Policy Challenge as well as 5 paper awards and honors. He has published more than 90 peer-reviewed papers in high impact journals and conferences. He has raised significant funding and worked for EU projects such as ASSET, SoBigData and FuturICT 2.0.
Title: Coordination of Drones at Scale in Smart Cities: From Distributed Optimization to Multi-agent Reinforcement Learning
Abstracts: Smart C
ions such as traffic monitoring and disaster response can be revolutionized by swarms of intelligent and cooperative drones. Drones can collect sensor data in parallel over different areas of interest and time spans within unpredictable environments. However when missions become spatio-temporally large and evolving during flying operations, recharging drones while meeting sensing requirements and battery constraints becomes a challenge. To address this timely problem we first introduce a scalable and energy-aware model for planning and coordination of drones for spatio-temporal sensing. The coordination model is built upon a decentralized multi-agent collective learning algorithm (EPOS) to ensure scalability, resilience and flexibility that existing approaches lack of. We then augment this novel approach with a multi-agent deep reinforcement learning approach to account for long-term strategic decisions made collectively by the drones to maximize the accumulated performance benefits. This allows swarms of drones to self-optimize their operations while they learn to “follow” the phenomenon they observe, e.g. traffic jams or a forest fire. Experimental and analytical results illustrate the outstanding performance of the proposed methods compared to state-of-the-art approaches, while providing a deeper understanding on coordinated mobility of drones.
Speaker: Prof. Evangelos Pournaras
Affiliations: University of Leeds, UK