Hands-On with Robots: Collaborating on Unitree Go2 at ELTE
In my previous post, I shared my travels across Hungary. But let's shift gears to something closer to my daily grind: interactive projects that bridge the classroom and the real world. For me, that means diving into robotics with industry partners like Bosch. It's one of those experiences that makes ELTE feel like a launchpad, not just a lecture hall.
It all started last spring during ELTE's annual "Dog Day" events, a fun, casual gathering where students and their pets (or robotic stand-ins) compete in an obstacle course. Picture this: real dogs navigating agility tunnels, jumps, and weaves alongside their owners, barking and wagging tails everywhere. Then, out of nowhere, a sleek robotic dog joins the fun, trotting smoothly through the course without missing a beat. I was stunned. "What's that?" I asked my professor, who was cheering on the human-dog teams. Turns out, it was the Unitree Go2 Edu, a quadruped robot designed for research and education. That sparked an instant conversation about AI in robotics, control systems, and how something like this could tie into my thesis on reinforcement learning.
From that casual chat, things snowballed. My professor connected me with a research team, who were already using Unitree platforms for autonomous navigation and perception projects. Bosch, with its global push into smart mobility and Industry 4.0, saw potential in ELTE's AI expertise for enhancing robot behaviors in dynamic environments. I jumped in as a student collaborator, working part-time on the Go2 Edu's development. It wasn't a formal internship, just hands-on prototyping, guided by ELTE labs in Bosch lab.
My role? Integrating AI modules for gesture recognition and basic RL-based obstacle avoidance. We'd start with simple tasks: training the robot to respond to hand signals using computer vision libraries like OpenCV and PyTorch.
I spent evenings tweaking models on my laptop, then testing them in the lab where the Go2 would "learn" to pause or turn based on detected gestures. I had to build everything from the ground up, starting with basic industrial simulations. Then we kept improving things step by step by working together through shared Git repositories - everything was organized, tracked, and easy for the team to collaborate on. One hilarious mishap: during a test run, the robot misinterpreted a thumbs-up as "full speed ahead," zooming straight into a foam barrier. We laughed it off, debugged the confidence thresholds, and improved the model. Those moments made the tech feel alive.
What I loved most was the blend of theory and practice. ELTE's AI courses gave me math, while Bosch's input grounded it in real-world challenges. The project even fed into my thesis, exploring how Reinforcement Learning (RL) can make robots more adaptive in unpredictable settings.
For fellow ELTE students: if you're eyeing robotics or AI, chat with professors at events like Dog Day gateways to these collabs. Bosch and similar partners (think Siemens or local startups) often seek fresh talent for quick prototypes. It's not always glamorous (late-night debugging is real), but it's rewarding. My Go2 work not only boosted my skills but also my confidence to tackle industry problems head-on.
If you're inspired, check ELTE's robotics labs or reach out to faculty. Who knows? Your next project could be the one that shapes your career.
Stay curious!
Thausif