Yann Epars

École Polytechnique Fédérale de Lausanne

Papers

Total Citations

241

H-Index

About

Yann Epars is a researcher whose work lies at the intersection of robotics, artificial intelligence, and cognitive modeling, with a particular focus on imitation learning and neural evolution. His most influential contribution, "Discovering optimal imitation strategies" (2004, 175 citations), addresses a fundamental challenge in robotics: determining *what* to imitate from a demonstration. Epars modeled the imitator's strategy as a hierarchical optimization system that categorizes multi-dimensional data to identify which features of a demonstration are task-relevant and should be reproduced. This work, alongside his earlier paper on the same topic (2003, 21 citations), provides a principled framework for robots to autonomously extract and replicate essential behaviors. Epars also made notable contributions to evolutionary robotics with his 2006 study (45 citations), where he demonstrated that simple genetic representations and fitness functions could rapidly evolve spiking neural circuits capable of controlling autonomous mobile robots in textured environments. While his citation counts reflect a focused, high-impact body of work rather than broad recognition, Epars' research offers foundational insights for students and researchers interested in how robots can learn from observation and how neural architectures can be evolved for complex sensorimotor tasks.