We went to the industry with one idea and came back with another

Gemm's first idea was spectacular: a humanoid that walks into a plant during a gas leak and closes the valve. We had the four-legged robot and were waiting for the humanoid. Before building more, we went to ask the people who send humans into those places.

The meetings

With YPF we had two video calls and one in-person meeting with their robotic inspection team, who referred us to Y-TEC in La Plata. With Pan American Energy we talked to operations. For each one we prepared concrete questions instead of a pitch: which rounds are done today just to look or listen, what information they'd like more often, where a robot would make sense and where it wouldn't.

What they told us

• —Safety is rule number one. Before anything else.
• —What's useful is frequent, not spectacular. Routine inspections, surveillance, spotting a forgotten tool or tall grass. Not rescues.
• —They already use drones. A ground robot wins where a drone can't go: under pipes, indoors, and when measuring gas, because drone propellers disperse the concentration and a robot doesn't.
• —Our robots aren't explosion-proof. Without ATEX certification they can't enter classified zones, so any test has to happen where a regular vehicle could drive.

What changed

We went from a robot that turns valves to a robot as a sensor platform: cameras, audio, gas and LiDAR walking the plant and warning before something happens. And we accepted something we didn't want to at first: autonomy isn't all or nothing. First a person operates the robot from somewhere safe; then you automate what already works.

Today Gemm offers exactly that as a pilot: pick a real round with the plant, map it, walk it with the teleoperated robot and compare data, times and exposure with the current round. Teleoperation is the piece that makes that first step possible, and it's my part.

What I take from it

• —Something people like in a presentation isn't something anyone needs.
• —Specific questions, not generic ones. "Which rounds do you walk just to look?" opened far more than "would a robot help you?".
• —These meetings came out of a school project. The network you build is worth as much as what you build.