Papers
29
Total Citations
1,044
H-Index
13
About
Roland Hafner is a prominent researcher specializing in reinforcement learning (RL) for robotics, with a particular focus on robot locomotion, dexterous manipulation, and autonomous skill acquisition. His career spans nearly two decades of foundational and applied contributions to the field, beginning with pioneering work on neural reinforcement learning controllers for real robots in 2007 and culminating in landmark achievements such as training bipedal humanoid robots to play soccer using deep RL (2024, 147 citations). Hafner's most influential contributions include the development of Scheduled Auxiliary Control (SAC-X), a novel learning paradigm enabling agents to master complex behaviors from sparse reward signals (2018, 155 citations), and data-efficient deep RL methods for dexterous manipulation (2017, 118 citations). His early work on reinforcement learning for robot soccer (2009, 266 citations) remains his most cited achievement, establishing him as a foundational voice in autonomous robot learning. He has also advanced offline RL through behavioral modeling priors and tackled hybrid discrete-continuous control challenges in robotics. Across his body of work, Hafner consistently bridges theoretical RL innovation with real-world robotic deployment, making his research invaluable for students and practitioners working at the intersection of machine learning and intelligent physical systems.
Research Focus
Key Achievements
Top Papers
- 1Reinforcement learning for robot soccer266 citations · 2009
- 2Learning by Playing - Solving Sparse Reward Tasks from Scratch155 citations · 2018
- 3
- 4Data-efficient Deep Reinforcement Learning for Dexterous Manipulation118 citations · 2017
- 5
- 6Neural Reinforcement Learning Controllers for a Real Robot Application48 citations · 2007
- 7
- 8Continuous-Discrete Reinforcement Learning for Hybrid Control in Robotics27 citations · 2020
- 9
- 10Making a Robot Learn to Play Soccer Using Reward and Punishment16 citations · 2007