The problem
A shared lab full of expensive machines has no memory. Anyone can walk in, use a lathe or a laser cutter, and walk out. When something breaks there is no record of who was on it, and no way to stop someone untrained from starting it in the first place.
What it does
Access control for physical machines, deployed across five of them.
- ESP32 + MFRC522 RFID nodes at each machine, with relay actuation on the machine’s power. No card, no power.
- An automated card dispenser — N20 motor with IR sensing — so issuing access does not require a person standing there.
- A Raspberry Pi 5 issuing station for registering users and provisioning cards.
- A Firebase Firestore backend with a real-time web dashboard: who is on what, right now, and who was on it before.
Every machine session is traceable to a person. Training requirements can be enforced at the relay rather than on a poster nobody reads.
How it is built
Each node is standalone: it makes its own allow/deny decision from a locally cached credential set, so a network drop does not take the workshop offline. State syncs back to Firestore when the connection returns.
The dispenser was the awkward part. Reliably feeding one card and exactly one card, with a cheap motor and an IR break-beam, took more revisions than the access logic did.
Where it went
- IITISoC Gold Medal, Robotics domain.
- Grand Innovation Award and Best Maker Award, Indian Academic Makerspaces Summit 2025, IIT Gandhinagar. ₹62,500.
- Patent in progress.
- Still running in Tinkerers’ Lab, IIT Indore. It is not a demo; it is infrastructure other people depend on, which means it has to keep working when nobody is looking at it.
It has since grown into a broader machine management system, and it is one of two projects that may turn into something larger.