Papers
25
Total Citations
614
H-Index
13
About
Markus Wulfmeier is a leading researcher at the intersection of deep reinforcement learning, robot learning, and autonomous systems, with a career spanning robotic locomotion, domain adaptation, and imitation learning. His most celebrated work includes pioneering efforts in training bipedal humanoid robots to play soccer using deep RL, demonstrating that sophisticated, agile movement skills can emerge in low-cost hardware — a paper that has already garnered 147 citations since 2024. Wulfmeier has made foundational contributions to curriculum-based reinforcement learning, with his Reverse Curriculum Generation framework (140 citations) offering an elegant solution to goal-oriented manipulation tasks that remain challenging for standard RL approaches. His research extends into sim-to-real transfer through mutual alignment and adversarial domain adaptation, addressing the persistent gap between simulated training environments and real-world deployment. Notable work on hybrid continuous-discrete control, hierarchical task decomposition via temporal alignment, and motion capture-driven locomotion skill reuse reflects the breadth of his contributions. More recently, Wulfmeier has explored foundation models as unified agents for robotics. From granular soil mechanics for lightweight vehicles to cutting-edge humanoid control, his trajectory illustrates a researcher consistently pushing the boundaries of what autonomous robots can learn and do.
Research Focus
Key Achievements
Top Papers
- 1
- 2Reverse Curriculum Generation for Reinforcement Learning140 citations · 2017
- 3
- 4TACO: Learning Task Decomposition via Temporal Alignment for Control29 citations · 2018
- 5Continuous-Discrete Reinforcement Learning for Hybrid Control in Robotics27 citations · 2020
- 6
- 7Mutual Alignment Transfer Learning20 citations · 2017
- 8
- 9
- 10Towards A Unified Agent with Foundation Models17 citations · 2023