Pieter Abbeel
University of California, Berkeley, International Computer Science Institute, University of California System, Technische Universität Darmstadt, Machine Intelligence Research Institute, OpenAI (United States), Stanford University, Carnegie Mellon University, Berkeley College, University of New Mexico, Rutgers, The State University of New Jersey, Baird Institute
Papers
271
Total Citations
36,280
H-Index
83
About
Pieter Abbeel stands as one of the most influential figures at the intersection of robotics, deep learning, and reinforcement learning. His research has fundamentally shaped how machines learn to perceive, plan, and act in complex real-world environments. Among his landmark contributions is Trust Region Policy Optimization (TRPO), a foundational reinforcement learning algorithm that introduced principled, monotonically improving policy updates and has accumulated over 3,100 citations, cementing its status as a cornerstone of modern RL. Alongside collaborators, he pioneered end-to-end deep visuomotor policy learning, enabling robots to acquire control skills directly from raw visual input without hand-engineered perception pipelines. His work on domain randomization (2,700+ citations) bridged the simulation-to-reality gap, while Soft Actor-Critic addressed critical challenges of sample efficiency and stability in continuous control. Beyond algorithms, Abbeel contributed practical tools for the robotics community, including the widely adopted YCB object benchmark and advances in motion planning. His research on DeepMimic further extended physically realistic character animation through reinforcement learning. With multiple papers surpassing 1,000 citations, Abbeel's cumulative influence across robotics manipulation, deep RL, and sim-to-real transfer has profoundly accelerated progress toward capable, autonomous robotic systems.
Research Focus
Key Achievements
Top Papers
- 1Trust Region Policy Optimization3,141 citations · 2015
- 2
- 3Soft Actor-Critic Algorithms and Applications1,952 citations · 2018
- 4High-Dimensional Continuous Control Using Generalized Advantage Estimation1,750 citations · 2015
- 5End-to-end training of deep visuomotor policies1,715 citations · 2016
- 6End-to-End Training of Deep Visuomotor Policies1,399 citations · 2015
- 7
- 8A Survey of Research on Cloud Robotics and Automation826 citations · 2015
- 9
- 10DeepMimic802 citations · 2018