In August 2024, Ibrahim Alispahić and Prof. Adnan Tahirović presented their research paper, Path Planning for Multi-Agent Systems Using Deep Q-Networks Reinforcement Learning, at the TAROS 2024 conference, held at Brunel University, London. This research, developed as part of Ibrahim’s thesis under the supervision of Prof. Tahirović at the Faculty of Electrical Engineering, University of Sarajevo, introduces an innovative approach to enhancing multi-agent cooperation and coordination using deep Q-networks (DQN).
The paper addresses the multi-agent traveling salesman problem, focusing on strategies to improve collaboration and efficient target-reaching—crucial for tasks like autonomous aerial inspections and maintenance. By applying the DQN-based algorithm, the research showcases the system’s flexibility in adapting to dynamic environments where both agents and targets change positions. This adaptability makes it highly applicable for real-world scenarios. Their work was presented during the Human-Robot Interaction/Collaboration and Multi-Agent Systems session, contributing to the broader discussion on advancements in autonomous systems and robotics fields. TAROS 2024 provided a valuable platform not only for sharing research but also for networking with researchers from leading UK universities. The event fostered meaningful discussions and opened the door to potential collaborations in the field of robotics and autonomous systems.