Robert Pinsler

Papers

Total Citations

H-Index

About

Robert Pinsler is a researcher whose work lies at the intersection of reinforcement learning, robotics, and human-in-the-loop systems. His key contributions focus on making robot learning more sample-efficient and feedback-efficient, particularly in complex, real-world environments. His most cited work, "Sample and Feedback Efficient Hierarchical Reinforcement Learning from Human Preferences" (2018, 19 citations), tackles the challenge of defining informative reward functions for physical robots. By integrating human preferences into a hierarchical reinforcement learning framework, Pinsler’s approach reduces the need for both extensive robot-generated samples and costly human feedback, enabling more practical and scalable robot training. This work is notable for bridging the gap between algorithmic efficiency and real-world deployment constraints. Additionally, in "Factored Contextual Policy Search with Bayesian Optimization" (2019, 4 citations), he addresses the problem of scarce data in complex tasks by developing a method that generalizes locally learned policies across different task contexts. This approach leverages Bayesian optimization to efficiently explore policy parameters, making it a valuable contribution to data-efficient robot learning. Pinsler’s research is particularly impactful for students and researchers seeking to deploy reinforcement learning in settings where data and human time are limited.