Fengyou Liu is an Electrical & Computer Engineering student at Princeton, class of 2029. He builds drones, robots, boats, and the electronics that run them. Welcome to his site.
I have a lot of projects, and most of them never get finished. The recent ones that did are documented here.
If you want to know more about any of them, please reach out!
An autonomous racing drone pilot. A flight stack built from the ground up, with vision, processing, sensing and RL policy for the 2026 Anduril AI Grand Prix competition. An ambitious project, making it through the first 2 qualifiers, but failing to land a top 10 spot for the in person competition. We'll get them next time. First clean lap: 38 seconds.
FIG. II ADROIT · the real thing · 2026 A two-legged robot that drives on powered wheels: legs for rough ground, wheels for speed. Its balance policy is a neural network trained by reinforcement learning in NVIDIA Isaac Lab, where it fell over a few thousand times so it wouldn’t have to in person, then deployed to run onboard a Jetson Orin Nano.
Roll the ball and the model rolls with it. A repurposed desk trackball that drives view navigation in Fusion 360, each axis of the ball mapped to an axis of the view, so one hand turns the part as if holding it while the other keeps sketching. A fraction of the price of the real thing.
FIG. III TRACKBALL · turning the view · 2026 A startup idea, still theoretical: a pick-and-place toolhead that bolts onto a common desktop 3D printer, borrowing its existing X-Y gantry instead of building new motion hardware. The printer already knows how to move precisely. Magpie teaches that same gantry to pick up components and set them down where they belong.
The most reliable way is email. I read everything.
Email is the best way to reach me: questions about a project, internships and research, or something you’re building yourself.
If you’re a student partway through your own build, I’m especially glad to hear from you.