Add Alessandro Pau abstract and photo

Author: jank324

assets/people/alessandro_pau.png

@@ -23,8 +23,12 @@ We have a very exciting line-up of keynote speakers, and are very much looking f
 Autonomous robots that can assist humans in situations of daily life have been a long-standing vision of robotics, artificial intelligence, and cognitive sciences. A first step towards this goal is to create robots that can learn tasks triggered by environmental context or higher level instruction. However, learning techniques have yet to live up to this promise as only few methods manage to scale to high-dimensional manipulator or humanoid robots. In this talk, we investigate a general framework suitable for learning motor skills in robotics which is based on the principles behind many analytical robotics approaches. To accomplish robot reinforcement learning from just few trials, the learning system can no longer explore all learn-able solutions but has to prioritize one solution over others – independent of the observed data. Such prioritization requires explicit or implicit assumptions, often called ‘induction biases’ in machine learning. Extrapolation to new robot learning tasks requires induction biases deeply rooted in general principles and domain knowledge from robotics, physics and control. Empirical evaluations on a several robot systems illustrate the effectiveness and applicability to learning control on an anthropomorphic robot arm. These robot motor skills range from toy examples (e.g., paddling a ball, ball-in-a-cup) to playing robot table tennis, juggling and manipulation of various objects.
 {{< /speaker >}}
 
-{{< speaker name="Alessandro Pau" position="Research Scientist" institute="Swiss Plasma Center - EPFL" image="people/placeholder.png" link="https://www.linkedin.com/in/alessandro-pau-a46916ba/?originalSubdomain=ch" >}}
-Nuclear fusion presents the potential to deliver a nearly unlimited energy source. One promising candidate for a future fusion reactor is the Tokamak, which magnetically encloses a plasma to achieve the extreme physical requirements. However, the plasma containment is prone to disruptions, hindering reliable operation. In this talk, we therefore explore learning-based solutions for plasma control and disruption prevention in Tokamaks.
+{{< speaker name="Alessandro Pau" position="Research Scientist" institute="Swiss Plasma Center - EPFL" image="people/alessandro_pau.png" link="https://www.linkedin.com/in/alessandro-pau-a46916ba/?originalSubdomain=ch" >}}
+Recent advances in Artificial Intelligence (AI), Machine Learning (ML), and Reinforcement Learning (RL) offer promising pathways toward addressing complex control problems in fusion plasmas and Tokamak devices. Tokamaks, which use magnetic fields to confine a plasma, present unique operational challenges due to their nonlinear dynamics, high-dimensional parameter spaces, and stringent real-time control requirements. In this keynote, we will first provide an accessible introduction to magnetic confinement and Tokamak operations, highlighting both the complexity of plasma physics and the critical control objectives necessary for sustaining stable fusion reactions.
+
+Subsequently, we will explore how traditional control methods can be integrated and enhanced by leveraging ML, AI, and particularly RL, to provide innovative solutions to these long-standing issues. Recent developments leveraging RL demonstrate significant potential for improved performance in key areas such as plasma shape control, stability and disruption prediction. We will present specific case studies and experimental outcomes, underscoring how RL algorithms navigate the complex interplay between plasma physics, actuator constraints, and control objectives.
+
+Finally, we will touch upon some current open challenges, including real-time implementation constraints, interpretability of learned policies, and scalability, while highlighting directions for future research.
 {{< /speaker >}}
 
 ## Impressions from RL4AA'24