Cem Eteke

Papers

Total Citations

H-Index

About

Cem Eteke is a researcher whose work sits at the intersection of robot learning and human-robot interaction, with a particular focus on making robotic skill acquisition more intuitive and efficient. His primary contributions lie in developing frameworks that enable robots to learn complex tasks from minimal human input. In his most cited work, "Reward Learning From Very Few Demonstrations" (2020, 18 citations), Eteke introduced a novel framework that learns reward functions from sparse demonstrations, using them to guide policy search for skill improvement. This approach, which models goals via hidden Markov models, significantly reduces the data burden on human teachers. Eteke has also explored the subtleties of non-verbal communication in "Communicative Cues for Reach-to-Grasp Motions: From Humans to Robots" (2018, 3 citations), where he conducted human-human experiments to identify the subtle cues that make intent communication fluent, with the goal of generating more expressive and predictable robot motion. His work is notable for bridging the gap between data-efficient learning and socially-aware robot behavior, offering practical pathways toward robots that can learn from and collaborate with humans more naturally.