Alex Irpan

About

Alex Irpan is a prominent robotics and machine learning researcher whose work has fundamentally shaped the intersection of deep reinforcement learning, large-scale foundation models, and robotic manipulation. Based at Google DeepMind, Irpan has made seminal contributions to vision-based robot learning, most notably through QT-Opt (2018, 575 citations), which demonstrated that scalable deep reinforcement learning could achieve robust robotic grasping from raw visual inputs. His research trajectory evolved impressively toward grounding language and world knowledge in physical robotic systems, contributing to landmark projects including "Do As I Can, Not As I Say" (2022, 516 citations) and the influential RT-1 and RT-2 Robotics Transformer series (collectively exceeding 800 citations), which showed that transformer architectures trained on large, diverse datasets could enable generalist real-world robot control. Irpan also advanced simulation-to-real transfer techniques through RL-CycleGAN and domain adaptation methods. His collaboration on Open X-Embodiment further cemented his role in building shared robotic learning infrastructure across institutions. With a cumulative citation count surpassing 2,400, Irpan's work continues to define how robots learn, generalize, and act intelligently in unstructured real-world environments.

Research Focus

Key Achievements

Top Papers

1. 1
   QT-Opt: Scalable Deep Reinforcement Learning for Vision-Based Robotic Manipulation

   575 citations · 2018
2. 2
   Do As I Can, Not As I Say: Grounding Language in Robotic Affordances

   516 citations · 2022
3. 3
   RT-1: Robotics Transformer for Real-World Control at Scale

   512 citations · 2023
4. 4
   RT-2: Vision-Language-Action Models Transfer Web Knowledge to Robotic Control

   267 citations · 2023
5. 5
   RL-CycleGAN: Reinforcement Learning Aware Simulation-to-Real

   154 citations · 2020
6. 6
   Open X-Embodiment: Robotic Learning Datasets and RT-X Models : Open X-Embodiment Collaboration⁰

   119 citations · 2024
7. 7
   Open X-Embodiment: Robotic Learning Datasets and RT-X Models

   101 citations · 2023
8. 8
   BC-Z: Zero-Shot Task Generalization with Robotic Imitation Learning

   89 citations · 2022
9. 9
   Using Simulation and Domain Adaptation to Improve Efficiency of Deep Robotic Grasping

   47 citations · 2018
10. 10
    RT-1: Robotics Transformer for Real-World Control at Scale

    38 citations · 2022

Key Collaborators

Contact & Links