Isabelle Sarda

About

Isabelle Sarda is a pioneer in the intersection of reinforcement learning and autonomous robotics, with a career dedicated to enabling machines to learn goal-directed behaviours through neural computation. Her foundational work, "Goal-directed behaviours by reinforcement learning" (1999), has garnered 7 citations and established a framework for training agents to pursue objectives without explicit programming. Earlier, in "Behaviour Learning by a Reward-Penalty Algorithm" (1995), she demonstrated how neural networks could master complex motor tasks—from gait learning to obstacle avoidance—using a simple reward-penalty mechanism, a contribution that has earned 5 citations and remains influential in adaptive control. Sarda’s research elegantly bridges theoretical reinforcement learning with practical robotic applications, showing how trial-and-error learning can produce robust, real-world behaviours. Her work is particularly notable for its early exploration of biologically inspired learning rules, predating the deep reinforcement learning revolution. For students and researchers, Sarda’s legacy offers a clear, principled path from algorithm design to embodied intelligence, proving that even modest citation counts can reflect profound, lasting impact on how machines learn to act in dynamic environments.

Research Focus

Key Achievements

Top Papers

1. 1
   Goal-directed behaviours by reinforcement learning

   7 citations · 1999
2. 2
   Behaviour Learning by a Reward-Penalty Algorithm : From Gait Learning to Obstacle Avoidance by Neural Networks

   5 citations · 1995

Key Collaborators

Contact & Links