Giulio Romualdi

Humanoid locomotion tech lead at Generative Bionics. I make humanoid robots walk — blending optimal control, MPC and learning.

Continue

こんにちは! · Hello!

I make humanoid robots walk.

I'm Giulio Romualdi, humanoid locomotion tech lead at Generative Bionics. I design the control and learning stack that turns whole-body dynamics into robust, human-like walking — from optimal control and model predictive control to reinforcement learning.

Previously I was a postdoctoral researcher and Scrum Master of the locomotion team at the Artificial and Mechanical Intelligence Lab, Istituto Italiano di Tecnologia, where I also earned my Ph.D. in Humanoid Robotics with the University of Genoa. I maintain the open-source Bipedal Locomotion Framework.

What's new

News & Highlights

Now · Generative Bionics
As product owner, I led the team that delivered Motion Intelligence via RL on the GENE.01 lower-body platforms — our three-month milestone bringing reinforcement-learning locomotion to multiple humanoid platforms, built on top of the Bipedal Locomotion Framework.
2024 · IEEE Humanoids
Best Oral Presentation Finalist for Online DNN-driven Nonlinear MPC for Stylistic Humanoid Robot Walking, demonstrating human-like walking with online step adjustment on the ergoCub robot.
2024 · Science Robotics
Co-authored the iCub3 Avatar System, enabling fully immersive remote embodiment of a humanoid robot across a 300 km link.
2023 · Public outreach
The ergoCub robot walked on national Italian television, showcasing the locomotion architecture developed by our team.

Selected work

Research Highlights

First-author projects on planning, control and learning for humanoid bipedal locomotion. Each has its own project page.

Humanoids 2024

DNN-driven Nonlinear MPC for Stylistic Walking

A three-layer architecture combining a deep neural network trajectory generator with model-predictive step adjustment for human-like, disturbance-robust walking.

Project page →MPC · Learning

ICRA 2022