Last Friday, 9 May, the seventeenth edition of the ValgrAI Research Matinees was held, consolidating itself as a key space for the exchange of knowledge in artificial intelligence (AI) in the Valencian Community. This monthly event, organised by the ValgrAI Foundation, brought together researchers, students and professionals to explore the latest advances in AI and foster interdisciplinary collaboration.
Robot control based on deep reinforcement learning
After reviewing classical control methods and their limitations in dealing with complex tasks or changing environments, Belmonte explained how reinforcement learning allows robots to learn through experience, maximising rewards without the need to explicitly model the environment. By incorporating deep neural networks, this approach allows robots to develop sophisticated and adaptive behaviours.
The talk included several examples of practical application in robotic manipulation, locomotion and coordination of multiple robotic arms. Special mention was made of the field of orbital robotics, where DRL is opening up new possibilities for autonomous control in space scenarios.
The importance of considering realistic aspects for the validation of ambulance relocation strategies
Yulia Karpova then presented an interesting research focused on improving response times in Emergency Health Services. Specifically, she addressed the issue of dynamic relocation of ambulances as an alternative to fleet expansion, a solution that has significant operational and economic advantages.
However, as the researcher explained, many of the strategies developed over the last thirty years have not been implemented in real-life contexts due to their lack of consideration of practical factors such as actual call rates, geographical distribution, staff shifts or traffic restrictions.
The paper presented an advanced algorithm that incorporates these realistic elements into decision-making for ambulance relocation. To demonstrate its feasibility, Karpova analysed the case study of the city of Valencia, showing how the proposed approach can improve service efficiency and increase coverage without the need to add more vehicles. The research reinforces the idea that the practical utility of AI models depends, to a large extent, on their connection to operational reality.
Invitation to the Next Edition
The event not only provided a platform for the presentation of impactful research but also offered an opportunity for participants to interact with the researchers during a Q&A session, delving deeper into the topics discussed and exploring future collaborations.
The next edition of the ValgrAI Research Morning is scheduled for the first Friday of next month. We invite all interested parties to join us for another enriching session that promises more stimulating discussions and discoveries in the field of artificial intelligence. Mark your calendars and get ready to be inspired by the endless possibilities that AI can offer!
