Jonathan Booher
Papers
3
Total Citations
133
H-Index
3
About
Jonathan Booher is a leading researcher in robotics, specializing in imitation learning and scalable data collection for robotic manipulation. His major contributions center on addressing the critical bottleneck of acquiring large, high-quality training datasets for robots. Booher pioneered the RoboTurk platform, a crowdsourcing system that enables non-expert humans to remotely teleoperate robots, dramatically accelerating the collection of demonstration data. His seminal 2018 paper on RoboTurk, with 81 citations, demonstrated how imitation learning can overcome the exploration and reward specification challenges of reinforcement learning. Building on this, his 2019 work (cumulatively over 50 citations) showed how RoboTurk could scale robot supervision to hundreds of hours, creating richly annotated datasets that rival those in computer vision. By leveraging human reasoning and dexterity, Booher’s research has made it feasible to train robots on diverse manipulation tasks without costly expert supervision. His work is foundational for advancing robotic skill learning, bridging the gap between small-scale lab studies and real-world deployment, and has been widely recognized for its impact on the field.
Research Focus
Key Achievements
Top Papers
- 1
- 2
- 3