Russ Tedrake

Timeline

1. 1999–2004

   Michigan and an MIT PhD

   After high school in a Detroit suburb, he earns a B.S.E. in computer engineering at the University of Michigan in 1999 and spends a summer at the Santa Fe Institute simulating networks of neurons. He completes an MIT PhD in electrical engineering and computer science in 2004, doing his research in Sebastian Seung's lab in the Department of Brain and Cognitive Sciences.

2. 2005

   Toddler learns to walk

   As a postdoctoral associate in MIT's Department of Brain and Cognitive Sciences, he presents Toddler, a robot whose learning software teaches it to walk in less than 20 minutes, or about 600 steps. The work appears in Science in February 2005, coauthored with Andy Ruina, Steven Collins and Martijn Wisse.

3. 2005–2010

   A glider that perches

   In 2005, he and Rick Cory start work on a glider that can land on a wire like a bird, and Cory is named the 2010 Boeing Engineering Student of the Year for the project. As an EECS assistant professor, he is chosen for the 2008 Microsoft New Faculty Fellowship Program.

4. 2012

   Tenure and LQR-trees

   MIT's EECS department promotes him to associate professor with tenure in July 2012, and he receives the 2012 Ruth and Joel Spira Teaching Award. His Robot Locomotion Group at CSAIL builds bipeds that walk, quadrupeds that run and jump, and aircraft that land on a perch. A key result of his work is linear-quadratic regulator (LQR) trees, a tool for control guarantees on systems with complicated nonlinear dynamics.

5. 2015

   Atlas at the DARPA challenge

   He leads Team MIT's entry in the DARPA Robotics Challenge. At the June 2015 event in Pomona, California, the team's Atlas humanoid falls while leaving a vehicle and breaks its right wrist, and the team switches the robot to its left arm to continue the run.

6. 2022

   An ICML outstanding paper

   With Hyung Ju Suh, Max Simchowitz and Kaiqing Zhang, he writes "Do Differentiable Simulators Give Better Policy Gradients?", an ICML 2022 Outstanding Paper. The paper shows that stiffness or discontinuities in physical systems can weaken first-order gradient estimates for reinforcement learning, and proposes an alpha-order estimator that combines exact gradients with zero-order methods.

7. 2023–2025

   Large behavior models at TRI

   In September 2023, while he is vice president of robotics research at Toyota Research Institute, TRI unveils a diffusion-policy approach that has taught robots more than 60 dexterous skills from demonstrations, a step toward large behavior models (LBMs). In August 2025, as TRI's senior vice president of LBMs, he co-leads the Boston Dynamics project in which a single LBM controls the entire Atlas humanoid. Drake, a C++ toolbox started by the Robot Locomotion Group, grows under core development led by TRI.

8. Since 2026

   Co-founding Walden Robotics

   TRI spins out Walden Robotics in January 2026, and he leads it as co-founder and CEO after 10 years as a TRI senior vice president. Walden emerges from stealth in July 2026 with $300 million in funding and a wheeled semi-humanoid that works at a Toyota plant in North America, a design he favors because factories already have safety cases for wheeled robots. He remains the Toyota Professor of EECS, Aeronautics and Astronautics, and Mechanical Engineering at MIT.