About

Hi, I’m Romain Desarzens

Today, robots are leaving research labs and entering messy, unpredictable environments. For these systems to matter, they must move with reliability, awareness, and physical intelligence. Motion planning and control, the bridge between perception and action, determine whether a robot behaves like a simple tool or something we can trust.

My work focuses on turning mathematical models into robust, real-world robotic behavior.

I began in industrial robotics at iFollow, later Movu Robotics, joining when the company was still a small early team. I helped build the core navigation capabilities of their AMRs and took ownership of key software components as the product and the team scaled. Working in this environment required initiative, fast iteration, and clear technical leadership.

I then moved into autonomous vehicles at Navya, developing motion planning and MPC-based control algorithms for shuttles navigating complex urban environments. This experience reinforced a simple truth: the world never behaves like the model, so controllers must absorb uncertainty without drama.

After Navya, I returned to industrial robotics. I redesigned and hardened Movu Robotics navigation stack for large-scale warehouse deployments and contributed to the processes needed to operate robots reliably in production environments. Production constraints force clarity. Every assumption must be justified. Every millisecond counts.

In parallel, I explored legged locomotion as a personal research track. I implemented LIPM and ZMP preview control, IK-based gait generation, and simulation tools for multi-step walking. Locomotion requires mastery of geometry, dynamics, and real-time optimization. It is where theory meets gravity.

These domains, autonomous driving, industrial AMRs, and legged locomotion, share the same core challenge: design algorithms that make physical systems move predictably under uncertainty.

Today, I focus on motion control, trajectory optimization, and real-time robotics systems for robots that must act quickly and safely in complex environments. My goal is to help shape the next generation of autonomous agents that are grounded in mathematics, aware of physical limits, and reliable enough for the real world.