Marthens Hakizimana
Mechatronic Systems Engineer, EIT — Physical AI | Robotic Systems Modelling & Control | PhD Student, SFU
Mechatronic systems engineer (EIT, EGBC) focused on system modeling and control: deriving the dynamics of robots and processes, designing and analysing controllers against them, and implementing them on real hardware down to embedded firmware and ROS 2. Currently modelling and controlling industrial robotic welding cells at BCIT and SFU, with LLM- and vision-guided control in the loop, alongside FEA and mechanical design work at SKC Engineering.
About Me
Hello, I'm Marthens - a Robotics and Control Engineer in Training, currently a PhD student in Mechatronic Systems Engineering at Simon Fraser University, with a research focus on Physical AI and robotic welding systems at the BCIT Centre of Welding Technologies and Metallurgy Research and SFU Motion and Power Electronics Control Lab.
My work sits at the intersection of system modeling, control, and artificial intelligence. I start from the physics: I derive the dynamics of the machine or the process (Lagrangian and Newton–Euler dynamics for a robot arm, heat transfer and arc models for a weld), design the controller against that model, and check its stability before it goes anywhere near hardware. That runs from the computed-torque control of an in-pipe welding arm I built for my MASc, to a six-controller study of the Fanuc CRX-10iA and a stability analysis of a delayed seam-tracking loop, to putting LLMs and VLMs into a live robotic welding cell.
My current focus is trustworthy Physical AI: putting language models inside a live robotic welding cell without letting an opaque model drive a robot unchecked. That pulls me toward a set of questions I care about more broadly, cybersecurity, AI ethics, and governance, how powerful but opaque systems can be deployed safely, securely, and accountably in the physical world.
What sets my approach apart is that I carry a controller the whole way, from model to metal. I build the model, design and tune the control law in simulation and Hardware-in-the-Loop, and then implement it myself: embedded firmware on DSPs and ARM MCUs, the drive stages and wiring, the PCBs, and the ROS 2 software on top. Knowing the hardware first-hand — actuator dynamics, sensor noise floors, communication latencies, computational limits — feeds straight back into the model and the controller, instead of being a surprise at commissioning.
Beyond the lab, I'm an athlete at heart. I love football, running, and hiking.