Matthias Heyrman

Matthias Heyrman

Research Engineer · ETH Zürich

About

I am a research engineer working on producing visual locomotion and manipulation policies, focusing on humanoid robots.

Experience

  • Building scalable RL training and deployment pipelines for humanoid locomotion and imitation control on the Fourier N1 within IsaacLab, supporting research across multiple ongoing humanoid projects.
  • Improved autonomous indoor-vehicle reliability through data-driven system identification, an Acados MPC-based velocity controller, and a Bayesian state-estimation library.
  • Reduced emergency-stop incidents via predictive safety classification; automated deployment testing with a ROS-based log-verification suite.
  • Designed a planning and perception pipeline for human-robot collaborative manipulation, enabling multi-camera hand-eye calibration and ArUco-marker-based object localisation with OpenCV and ROS.
  • Mentored undergraduate research teams in hardware-software integration for bipedal locomotion projects, guiding students from concept through testing and iteration.
  • Contributed to the Cassie bipedal robot project across simulation tooling (Drake-based validation stack), terrain-sensing foot hardware design (actuated tarsal segments), and a drone–biped warehouse coordination architecture in C.

Skills

Programming Languages
Python C++ MATLAB C# C Java
Libraries & Packages
PyTorch NumPy SciPy ROS OpenCV Eigen Git
Simulation Tools
IsaacLab MuJoCo Gazebo
Human Languages
English — native French — native Dutch — basic Spanish — basic
Methods
Machine Learning Reinforcement Learning Model Predictive Control Recursive Estimation Dynamic Programming Computer Vision SLAM
Hobbies
Bouldering Hiking Photography Cooking

Education

  • Thesis: Multi-Domain Motion Embedding: Expressive Real-Time Mimicry for Legged Robots submitted Dec 2025