Information-theoretic complexity measures for evolutionary swarm robotics
Namur Institute for Complex Systems
This project aims to develop a novel information-theoretic framework for supporting the automated design of robot swarms, building on the principles of empowerment and relevant information, task-independent, universal and generic utility functions. The project will explore the potential of empowerment, which characterises the level of perceivable control an embodied agent has over its environment, and relevant information, which quantifies the minimal amount of information an agent needs to process in order to achieve a certain level of utility, for providing complexity measures of individual agents, swarms and the environment. By way of creating analytic models and artificial simulations, it will leverage the empowerment measure as a fitness function in swarm evolution and as a control mechanism in robot collectives. This will offer a novel perspective for evolutionary swarm robotics, building on generative mathematical models, objective quantitative measures and analytical tools.
My background is in Computer Science (Eötvös Loránd University, Hungary) and my research interests are focused at the intersection of Human-Computer Interaction, Artificial Intelligence and Complex Systems.
During my PhD studies (University of Glasgow, UK) I specialized in probabilistic information-theoretic modelling and analysis of computational and interactive cognitive systems. The work entailed the creation of generative models of human behaviour based on sensorimotor perception–action loops, as well as building machine learning models on experimental data collected in user studies.
As a research fellow at the Adaptive Systems Research Group (University of Hertfordshire, UK) I expanded my research horizon towards complex systems and artificial life in the context of social and human–robot interaction. In order to elucidate the fundamental information processing principles driving decision-making in living organisms I have developed information-theoretic models and tools for the study of human sensorimotor dynamics, robotic and simulated systems, based on behavioural and physiological sensing and analysis.
Previously, I have conducted research in the field of multimodal human-computer interaction (Nokia Research, Finland), exploring natural language, gesture and haptics interactive systems on mobile and embedded devices.
Most recently, as a post-doc at the Institute of Pervasive Computing (Johannes Kepler University, Austria), I have been involved in the development of embedded IoT devices for Industry 4.0, building on TinyML models, and have been teaching under- and postgraduate courses in Computer Science and AI, supervised and coordinated projects and internships.
D. Trendafilov, D. Polani, and A. Ferscha. Information-theoretic cost of decision-making in joint action. In Ana Paula Rocha ; Luc Steels and Jaap van den Herik: Proceedings of the 13th International Conference on Agents and Artificial Intelligence, 2021
D. Trendafilov, G. Schmitz, TH Hwang, AO Effenberg, and D. Polani. Tilting together: An information-theoretic characterization of behavioral roles in rhythmic dyadic interaction. Frontiers in Human Neuroscience, 14:185, 2020
D. Trendafilov and Daniel Polani. Information-theoretic Sensorimotor Foundations of Fitts’ Law. In Extended Abstracts of the 2019 CHI Conference on Human Factors in Computing Systems (CHI EA ’19), LBW2516, 1–6, 2019
D. Trendafilov, K. Zia, A. Ferscha, A. Abbas, B. Azadi, J. Selymes, M. Haslgrübler. Cognitive Products: System Architecture and Operational Principles. COGNITIVE, 2019
A. Maye, D. Trendafilov, D. Polani, and A. K. Engel, A visual attention mechanism for autonomous robots controlled by sensorimotor contingencies. International Conference on Intelligent Robots and Systems (IROS), 2015
D. Trendafilov, D. Polani, and R. Murray-Smith. Model of coordination flow in remote collaborative interaction. In UKSim International Conference on Modelling and Simulation, 2015
D. Trendafilov, R. Murray-Smith, and D. Polani. Empowerment as a metric for optimization in HCI. In CHI Workshop on Principles, Techniques and Perspectives on Optimization and HCI, 2015
D. Trendafilov, S. Lemmelä, and R. Murray-Smith. Negotiation models for mobile tactile interaction. In MSSP, Springer LNCS 8045, pp. 64–73, 2014
D. Trendafilov and R. Murray-Smith. Information-theoretic characterization of uncertainty in manual control. In Proceedings of IEEE SMC, 2013
K. Salminen, V. Surakka, J. Lylykangas, J. Rantala, T. Ahmaniemi, R. Raisamo, D. Trendafilov, and J. Kildal. Tactile modulation of emotional speech samples. Advances in Human-Computer Interaction, 2012
D. Trendafilov, Y. Vazquez-Alvarez, S. Lemmelä, and R. Murray-Smith. Can we work this out?: An evaluation of remote collaborative interaction in a mobile shared environment. In Proceedings of MobileHCI, 2011
E. Hoggan, D. Trendafilov, T. Ahmaniemi, and R. Raisamo. Squeeze vs. tilt: A comparative study using continuous tactile feedback. In Proceedings of ACM CHI Extended Abstracts, 2011
S. Lemmelä, A. Vetek, K. Mäkelä, and D. Trendafilov. Designing and evaluating multimodal interaction for mobile contexts. In Proceedings of ACM ICMI, 2008