Physical AI Hackathon for the Factory Floor
Four weeks. Real robots. Real factory problems.
Join the Physical AI Hackathon for the Factory Floor and teach a real production line to see, think, and act using AI agents and real robot arms. You bring the curiosity. We’ll bring the robots.
Register nowNo coding experience required. Spots are limited, so register early.
The biggest hackathon UWM has ever run
For four weeks, you’ll work on a real automated production line, the kind that mixes, fills, caps, and inspects products, with real FANUC robot arms, live cameras, and AI agents you can put to work. Your job is to make that line smarter: help it catch its own mistakes, explain what went wrong, or pick and place parts on its own.
You’ll build and test your ideas in a digital twin, a full simulation of the real equipment, then run them on the actual robots to see if they hold up. Closing that gap between “works in simulation” and “works in real life” is the whole game, and it’s exactly what NVIDIA and Rockwell engineers do every day.
This is the first time UWM students get their hands on this line, at the Connected Systems Institute, where we work with the same industrial AI tools companies bring to us every day.
Pick your challenge
Anyone can make a demo work on a screen. The hard part, and the part companies actually pay for, is proving it works on real equipment. Every challenge ends the same way: you run it on the real line and show it holds up.
See it: catch the defect
Build the AI eyes for the line’s inspection station. Spot a missing cap, a wrong fill level, or color that’s just starting to drift, and have an agent tell the operator what’s wrong and what to do next. The stretch goal: flag a problem while everything is still in spec, so the fix happens before the line ever goes down.
Understand it: find out why
Connect what the cameras see to the line’s own data and explain the root cause, not just the symptom. Real example: an orange batch follows a blue one, leftover color turns it an ugly green, and your system catches it, calls for a flush, and sends the bad product to scrap on its own. Detective work with real production data.
Act on it: teach the robot
Work in a digital twin of the FANUC arms, 3D print your own gripper fingers, and design your own objects to pick and sort. Tell an agent to pick up the yellow cubes and watch it find them, plan the move, and place them, first in simulation, then on the real arm.
You’ll work with NVIDIA’s simulation and AI tools, Rockwell’s control software, and real FANUC robot arms. Final challenge details and how you’ll be scored come at kickoff.
What you get out of it
- Work on real robots. Not a screen simulation you throw away. Real industrial arms, real equipment, a real line.
- The skills every job screens for. You’ll scope a messy problem, split the work across a team, build something that has to work on real hardware, and pitch it to technical and nontechnical judges. That’s the job, at every job.
- Get in front of NVIDIA and Rockwell. Their engineers are in the room mentoring and judging. Four weeks to show two major companies what you can do, with senior people watching.
- Cash prizes, plus GPUs, hardware, and compute from NVIDIA. Real awards for the strongest teams.
- A portfolio piece no classroom gives you. Physical AI on real hardware is exactly the kind of work that starts job conversations.
- No coding required. Come to learn. You’ll get AI agents to work with and people to help you get started.
Who should show up
- Open to all UWM students. It’s built for Computer Science, Data Science, and Mechanical, Electrical, Computer, and Industrial Engineering, but if you like solving real problems, you belong here.
- You don’t need to be a coder. You’ll work with AI agents, so curiosity and persistence matter more than your background or your GPA.
- Built to fit around your classes. The kickoff and finals are the only two must-attend dates. Everything in between runs on your and your team’s schedule, around your classes and other commitments, with optional hands-on help sessions whenever you want them.
- Every kind of builder has a role. Someone designs the gripper, someone wrangles the data, someone reads the sensors, someone tells the story to the judges. The best teams mix all of it.
- Solo or with a team. Teams are 4 or 5 people. Come with your own, or sign up solo and we’ll place you on one.
Cash, GPUs, and compute on the line
Full prize details announced soon. And the visibility with NVIDIA and Rockwell? That can be worth more than any prize.
Two dates to hold
These are your only must-attend moments. Everything in between is designed to fit around your classes and other commitments.
Kickoff at Rockwell Automation’s Milwaukee HQ
Finals at the Connected Systems Institute
Curious? Come eat some pizza.
Swing by CSI on September 29 from 4:00 to 6:00pm for free pizza and a look at the real line you’ll be working on. Meet the organizers, watch the robots run, and get your questions answered. Walk in and we’ll get you registered on the spot.
Questions, answered
Do I need to know how to code?
No. You’ll work with AI agents, and we’ll help you get started. Curiosity and persistence matter more.
Can I do this with a busy schedule?
Yes. It’s team-based and self-paced. The must-attend moments are the October 8 kickoff and the November 5 finals, plus optional hands-on sessions whenever you want help.
What will I actually work on?
You’ll pick a challenge on a real production line: catching defects, finding the root cause of a bad batch, or teaching a robot to grasp and sort. Final details come at kickoff.
Do I need a team?
No. Teams are 4 or 5 people. Bring your own, or sign up solo and we’ll place you on a team.
What does it cost?
Nothing. It’s free, and there’s free food.
Ready to build?
Register nowNo coding experience required. Spots are limited, so register early.
Questions? Come to the pizza party on September 29 from 4:00 to 6:00pm, or email Tom Hershberger at hershber@uwm.edu.