Papers
13
Total Citations
380
H-Index
8
About
Alex Bewley is a robotics and machine learning researcher whose work spans robot learning, domain adaptation, human-robot interaction, and high-speed robotic control. He has made significant contributions to the challenge of deploying learned models in real-world environments, particularly addressing how robots can remain robust to appearance changes caused by weather and seasonal variation through adversarial domain adaptation techniques, work that has accumulated over 80 citations across related publications. Bewley's most prominent contributions include his involvement in the landmark Open X-Embodiment project (220+ combined citations), a large-scale collaborative effort to consolidate diverse robotic learning datasets and train general-purpose robot foundation models — a milestone analogous to the pretrained model revolution in NLP and computer vision. His research also extends to human-aware robot navigation, leveraging human pose estimation for trajectory prediction, and sim-to-real reinforcement learning in tight human-robot interaction loops. Perhaps his most striking achievement is leading research on competitive robot table tennis, culminating in the 2025 paper "Achieving Human Level Competitive Robot Table Tennis" — the first learned robotic agent to reach amateur human-level performance in a physically demanding, real-time sport. Bewley's body of work reflects a consistent drive to push robotic systems from controlled laboratory settings into genuinely capable, real-world performance.
Research Focus
Key Achievements
Top Papers
- 1
- 2Open X-Embodiment: Robotic Learning Datasets and RT-X Models101 citations · 2023
- 3
- 4Robots That Can See: Leveraging Human Pose for Trajectory Prediction25 citations · 2023
- 5Robotic Table Tennis: A Case Study into a High Speed Learning System17 citations · 2023
- 6
- 7
- 8
- 9Large-scale outdoor scene reconstruction and correction with vision5 citations · 2020
- 10Achieving Human Level Competitive Robot Table Tennis3 citations · 2025