Researchers from the UDC and the CNR attended the IMOL Workshop in London to present the poster entitled: “A Motivational Architecture for Open-Ended Learning Challenges in Robots”. The article is written by Alejandro Romero, Gianluca Baldassarre, Richard J Duro, and Vieri Giuliano Santucci.

The International Conference on Intrinsically Motivated Open-Ended Learning (IMOL 2025) concluded at the University of Hertfordshire after three days of vibrant debate and cutting-edge presentations on autonomous robots.

The event was organised by 16 researchers, including Gianluca Baldassarre and Vieri Giuliano Santucci, both from the National Research Council of Italy, Institute of Cognitive Sciences and Technologies (ISTC-CNR). 

A highlight of the conference was the keynote address by Gianluca Baldassarre, who explored the motivational mechanisms underlying open-ended learning in robots, the talk named “Open‑Ended Learning Robots: From the Purpose Alignment Framework to a Purpose-Biased Model“. Baldassarre illustrated how to keep robots’ open learning in line with human goals and values. He introduced a goal-based framework that defines what a robot should learn, do, or avoid, regardless of the specific task. The framework specifies the conditions for alignment and breaks down the challenge into four areas:

• aligning human and robotic goals,
• managing multiple goals,
• translating abstract goals into concrete goals,
• acquiring the necessary skills.

The second part presents purpose-oriented open learning (POEL), which uses natural language, reasoning based on large language models, and computer vision to identify objects relevant to the purpose and guide the robot’s exploration toward meaningful interactions, while preserving open learning. Experiments show that POEL accelerates learning and improves performance compared to other methods, demonstrating that putting “purpose in the loop” allows autonomous robots to remain functional and safe for people.

The interesting Poster Session 2 presented innovative research, including:

• “A Motivational Architecture for Open-Ended Learning Challenges in Robots,” by Alejandro Romero, Gianluca Baldassarre, Richard J. Duro, and Vieri Giuliano Santucci, proposes a motivational framework to support continuous, adaptive robot learning.
  
• “Empowerment for Goal-driven Open-Ended Learning,” by Nicola Catenacci Volpi, Karen Archer, Emil Dmitruk, Kenzo Clauw, Frederick Richard, Emilio Cartoni, and Gianluca Baldassarre, presents empowerment as a key principle for autonomous, goal-oriented development.

With high-level speakers, interactive sessions, and strong international participation, IMOL 2025 confirmed itself as a milestone event for researchers shaping the future of cognitive robotics and open-ended artificial intelligence.