Erwin Coumans

About

Erwin Coumans is a pioneering researcher at the intersection of physics simulation, robotics, and machine learning, best known for his foundational work on sim-to-real transfer — the challenge of training robot behaviors in simulation and deploying them reliably in the physical world. His most influential contribution, "Sim-to-Real: Learning Agile Locomotion For Quadruped Robots" (2018), has accumulated nearly 800 citations across multiple venues, demonstrating how deep reinforcement learning can automate the design of complex legged robot controllers from scratch. Coumans has consistently advanced the frontier of agile robotic locomotion, exploring meta-learning for rapid adaptation, animal-motion imitation for naturalistic movement, and modular policy architectures that blend learned behaviors with structured trajectory generators. His work extends beyond locomotion into robotic manipulation of deformable objects and high-speed systems like table tennis, showcasing remarkable breadth. He is also the creator of the widely used Bullet Physics engine and PyBullet simulator, tools that have quietly underpinned countless robotics and AI research projects worldwide. With hundreds of citations spanning foundational and applied work, Coumans has shaped how the modern robotics community approaches learning-based control.

Research Focus

Key Achievements

Top Papers

1. 1
   Sim-to-Real: Learning Agile Locomotion For Quadruped Robots

   673 citations · 2018
2. 2
   Learning to Rearrange Deformable Cables, Fabrics, and Bags with Goal-Conditioned Transporter Networks

   122 citations · 2021
3. 3
   Sim-to-Real: Learning Agile Locomotion For Quadruped Robots

   114 citations · 2018
4. 4
   Learning Fast Adaptation With Meta Strategy Optimization

   84 citations · 2020
5. 5
   Learning Agile Robotic Locomotion Skills by Imitating Animals

   41 citations · 2020
6. 6
   Policies Modulating Trajectory Generators

   38 citations · 2019
7. 7
   Learning Agile Robotic Locomotion Skills by Imitating Animals

   34 citations · 2020
8. 8
   Robotic Table Tennis: A Case Study into a High Speed Learning System

   17 citations · 2023
9. 9
   NeuralSim: Augmenting Differentiable Simulators with Neural Networks

   13 citations · 2021
10. 10
    Multi-Task Learning with Sequence-Conditioned Transporter Networks

    13 citations · 2022

Key Collaborators

Contact & Links