Papers
8
Total Citations
1,035
H-Index
8
About
Nate Kohl is a pioneering researcher in machine learning for robotics and autonomous systems, best known for his groundbreaking work in applying neuroevolution to real-world control problems. His primary research areas include reinforcement learning for legged locomotion, evolutionary robotics, and multi-agent systems. Kohl's most influential contribution is his 2004 paper on policy gradient reinforcement learning for fast quadrupedal locomotion, which has accumulated over 585 citations. This work demonstrated how machine learning could automatically optimize a trot gait for forward speed on physical robots, marking a significant advance in legged locomotion. His follow-up paper on machine learning for fast quadrupedal locomotion (151 citations) further established his reputation in the field. Kohl also made notable contributions to multi-agent systems through his work on evolving keepaway soccer players using task decomposition, and extended neuroevolution techniques to practical applications like vehicle warning systems. His research on autonomous learning of stable quadruped locomotion and integrated neuroevolutionary approaches to reactive control and high-level strategy has influenced subsequent work in evolutionary robotics and adaptive control systems.
Research Focus
Key Achievements
Top Papers
- 1Policy gradient reinforcement learning for fast quadrupedal locomotion585 citations · 2004
- 2Machine learning for fast quadrupedal locomotion151 citations · 2004
- 3Evolving Soccer Keepaway Players Through Task Decomposition95 citations · 2005
- 4Evolving Keepaway Soccer Players through Task Decomposition76 citations · 2003
- 5Evolving a real-world vehicle warning system52 citations · 2006
- 6Autonomous Learning of Stable Quadruped Locomotion37 citations · 2007
- 7From pixels to multi-robot decision-making: A study in uncertainty21 citations · 2006
- 8