01About
Real work takes the whole body.
A humanoid robot is useful when it can move, keep its balance and use its hands at the same time. I develop AI for this whole-body motion, from planning how the robot moves to the learned control that carries it out.
My first humanoid was TEO, a full-size humanoid at UC3M in Madrid, where I started planning its movements in 2013. My PhD at IIT then studied how a robot can combine skills, learn from failures as well as successes, and pass what it knows to a different robot. Nearly seven years on legged robots followed, most at ANYbotics, where learned locomotion had to hold up on a commercial robot. At Hexagon Robotics I bring that experience back to humanoids, on AEON.
02Research
Reinforcement learning at the core.
-
01
Hierarchical and multi-task policies
Policies trained with deep reinforcement learning in massively parallel simulation. My focus is their structure: policies that handle many tasks and objectives at once, and hierarchies that build new skills from the ones a robot already has.
-
02
Whole-body humanoid motion
Humanoid motion, from trajectory generation to learned control on the robot. Next, I want to explore world models and vision-language-action models for whole-body behavior.
-
03
Perceptive locomotion and navigation
Learned controllers with multi-modal inputs, combining the robot’s sense of its own body with what it perceives of its surroundings, to move and navigate robustly through real environments. I put as much care into the training pipelines, benchmarking and validation that measure the gap between simulation and the real robot, instead of assuming it is closed.
03Experience
Research labs and industry.
-
2026 – now
Senior AI & Robotics Engineer
Hexagon RoboticsZurich
Whole-body robot learning for AEON, a wheeled humanoid.
-
2021 – 2026
Senior AI & Robotics Engineer
ANYboticsZurich
RL locomotion for ANYmal, from training to real-world deployment.
-
2019 – 2021
Postdoctoral Researcher
Dynamic Legged Systems Lab, IITGenoa
Vision-based and RL locomotion for legged robots.
-
2015 – 2019
PhD in Robotics
Learning and Interaction Group, IITGenoa
Multi-task reinforcement learning and skill transfer for humanoids.
04Contact
Say hello.
For roles, talks or collaborations, LinkedIn is the best way to reach me.