Towards Automated Algorithm Design: A Case Study in Swarm Robotics
Robot swarms, where multiple autonomous agents collaborate to achieve collective goals, are gaining attention for their scalability, resilience, and adaptability across domains like logistics or agriculture. Swarm intelligence, inspired by self-organizing behaviors in nature, provides a powerful framework for managing these systems. However, designing effective swarming algorithms remains a significant challenge due to the complexity of agent interactions and the multi-objective nature of the problem.
This presentation traces the evolution of swarming algorithm design, from manual, nature-inspired approaches to advanced automated methods. The design of novel machine learning algorithms is nowadays enabling the generation of efficient, reusable, and novel swarm behaviors, significantly reducing design costs and human intervention. This talk will explore how this shift towards automated algorithm design is transforming the development of robot swarms, with a focus on aerospace applications.
Grégoire Danoy holds a Ph.D. in Computer Science (2008) from the École des Mines de Saint-Étienne, France. He is currently a Research Scientist at the Interdisciplinary Centre for Security, Reliability, and Trust (SnT) of University of Luxembourg where he is also the Head of the Parallel Computing and Optimization Group (PCOG). He serves on the Editorial Board of Engineering Applications of Artificial Intelligence (EAAI).
His research focuses on designing novel artificial intelligence techniques, with an emphasis on optimization, swarm intelligence, and machine learning, with applications in unmanned autonomous systems (e.g., swarms of drones), cloud computing, high-performance computing, and smart mobility. He has authored more than 150 research articles in leading international journals and conferences.