Vladlen Koltun

About

Vladlen Koltun is a pioneering researcher whose work spans robotics, autonomous systems, and artificial intelligence, with particular expertise in reinforcement learning for agile robotic control. His most celebrated contributions have reshaped how machines learn to move and navigate complex environments. His 2019 work on legged robot locomotion (1,398 citations) demonstrated that reinforcement learning could surpass hand-crafted control methods, enabling animals-like agility in robotic systems — a breakthrough that continues to influence the field. Building on this, his 2022 research on perceptive quadrupedal locomotion (729 citations) pushed robots toward real-world deployment in hazardous terrain. Perhaps most dramatically, his team achieved champion-level drone racing through deep reinforcement learning (562 citations), defeating human professionals in a high-speed competitive setting. Koltun has also made foundational contributions to autonomous driving, developing conditional imitation learning frameworks (1,065 citations) and sim-to-real transfer techniques. His Habitat platform (197 citations) has become an important infrastructure for embodied AI research. Spanning computational geometry, acoustic imaging, and optimal control, Koltun's remarkably broad portfolio reflects both rigorous theoretical grounding and transformative real-world impact, making him one of the most influential figures in modern robotics and AI.

Research Focus

Key Achievements

Top Papers

1. 1
   Learning agile and dynamic motor skills for legged robots

   1,398 citations · 2019
2. 2
   End-to-end driving via conditional imitation learning

   1,065 citations
3. 3
   Learning robust perceptive locomotion for quadrupedal robots in the wild

   729 citations · 2022
4. 4
   Champion-level drone racing using deep reinforcement learning

   562 citations · 2023
5. 5
   Habitat: A Platform for Embodied AI Research

   197 citations · 2019
6. 6
   Reaching the limit in autonomous racing: Optimal control versus reinforcement learning

   194 citations · 2023
7. 7
   Driving Policy Transfer via Modularity and Abstraction

   122 citations · 2018
8. 8
   Acoustic Non-Line-Of-Sight Imaging

   102 citations · 2019
9. 9
   Trajectory Optimization With Implicit Hard Contacts

   69 citations · 2018
10. 10
    Almost tight upper bounds for vertical decompositions in four dimensions

    59 citations · 2004

Key Collaborators

Contact & Links