Tom Erez

About

Tom Erez is a prominent researcher at the intersection of robotics, physics simulation, and reinforcement learning, whose work has profoundly shaped how intelligent systems learn and execute complex motor behaviors. Best known as a key contributor to the MuJoCo physics engine, Erez co-authored a landmark comparative study of robotics simulators (314 citations) that helped standardize simulation practices across the field. His 2012 paper on online trajectory optimization (760 citations) demonstrated that humanoid robots could synthesize acrobatic, recovery-oriented behaviors in real time — a milestone in model predictive control. Building on this foundation, he developed integrated systems for real-time humanoid control and pioneered early work in Differential Dynamic Programming (117 citations) for high-dimensional nonlinear systems. Erez also made significant contributions to deep reinforcement learning, co-developing the widely adopted dm_control software suite (186 citations) and advancing data-efficient and imitation-assisted learning for dexterous robotic manipulation. His "Catch & Carry" work further extended these ideas to full-body physics-based character animation. With thousands of citations across robotics, machine learning, and animation, Erez's research has become essential reading for anyone working on physically grounded intelligent agents.

Research Focus

Key Achievements

Top Papers

1. 1
   Synthesis and stabilization of complex behaviors through online trajectory optimization

   760 citations · 2012
2. 2
   Simulation tools for model-based robotics: Comparison of Bullet, Havok, MuJoCo, ODE and PhysX

   314 citations · 2015
3. 3
   Reinforcement and Imitation Learning for Diverse Visuomotor Skills

   217 citations · 2018
4. 4
   dm_control: Software and tasks for continuous control

   186 citations · 2020
5. 5
   An integrated system for real-time model predictive control of humanoid robots

   126 citations · 2013
6. 6
   Data-efficient Deep Reinforcement Learning for Dexterous Manipulation

   118 citations · 2017
7. 7
   Receding Horizon Differential Dynamic Programming

   117 citations · 2007
8. 8
   Reinforcement and Imitation Learning for Diverse Visuomotor Skills

   116 citations · 2018
9. 9
   Catch & Carry

   98 citations · 2020
10. 10
    Real-time behaviour synthesis for dynamic hand-manipulation

    65 citations · 2014

Key Collaborators

Contact & Links