Interview by: Marianne Santoro
C. 2024, All Rights Reserved
For Liberty Robotics engineers and technicians, reliability isn’t just an afterthought. It’s built into everything we do from the ground up. If the machine vision hardware can’t survive the environment into which you install it, you’re out of luck. Our hardware is built to work in the most challenging factory environments on earth.
I sat down and had a conversation with Liberty Robotics CTO, G Neil Haven, about just what goes into our robust reliability.
“We’ve learned a lot about materials handling from our over 10 years experience with the automotive industry. The factory environment in automotive is as harsh as it gets anywhere.”
Robust reliability starts with the hardware.
“We’ve installed our systems into everything including unheated, un-air conditioned plants in Japan and around the Mason Dixon line in the United States, all the way up to modern climate controlled factories.”
Sensor design counts.
“One part is the sensor design. Our sensor design is not a consumer grade sensor. It’s an industrial grade sensor. We have it packaged so that it can take a beating and keep on working.
Our sensor is in a metallic enclosure designed with heat dissipation in mind. It’s designed so that, for instance, if the front plate gets abraded, or somebody sprays paint on it, we can just replace the front plate without replacing the entire sensor.”
A single connector cable for everything.
“There’s only one connector, and it’s an industrial grade connector that goes from our sensor back to the computing hardware. And that one connection carries both power and signal. Another important thing is that it’s not a USB connector.
It’s something called a T12. It carries a GigE signal and it also carries power at the same time so the whole thing only needs one cable. We designed the wiring, so that it can be mounted on a robot and flexed, I think 10 million times is the spec. A lot of our competitors have USB connectors that may only flex a few thousand times before they may have to be replaced.”
Industrialized hardware is, well, harder.
“Our computer hardware is an industrial PC mounted in an industrial controls cabinet. So the whole thing is industrialized, and this is something that we’re carrying over from the automotive market into the warehousing market. A lot of our competitors that we see show up with just consumer grade sensors, poor connectivity, and no special thought paid to their computer hardware, while we thought through all of that stuff.”
Our reliability was starting to make a lot of sense.
I asked him, “So we’re showing up to the varsity match with varsity grade equipment, whereas a lot of our competitors are still showing up ready for JV?”
“Abso-friggin-lutely,” was Neil’s response.
He went on, “In particular, you’ll see our competitors use plastic housed sensors that maybe they’ve put some box around, or maybe they haven’t. Our stuff is ready to be robot-mounted or static-mounted, and it’s designed without any moving parts.”
“So you could probably put it in a big basketball for a while, play a game with it, and afterwards it would still work?” I asked.
“There aren’t any moving parts, so that might work. That would make an interesting experiment.”
“And a great commercial,” I added.
Test everything for anything that could possibly go wrong.
“Yeah, we’ve put them through vibration testing and electrical noise testing. We’ve zapped them with static changes, just zapped them. And we’ve also put them in an environment with big electrical fields to try to induce electromotive or electrical interference.
The reason we had to go through that EMI testing is that robots have these motors mounted on them, and when the motors move, they draw lots of current. When they draw that current, that induces electrical and magnetic fields that are very strong.
If our sensor is mounted on the end-arm of a robot, and if our signal cable runs along the arm of the robot in the vicinity of these motors, then we have to be immune to not just ambient light but ambient electromagnetic interference as well.
We just made sure that all that stuff works.
If you want dancing humanoid robots, don’t come to us. But if you want materials handling robots with a less than 1 in 1 million fail rate, give us a call.”
I had wondered exactly why we were so far ahead of the competition, but now the reasons seemed obvious and even simple. “Numbers like that are understandable, when you know what goes into them, aren’t they?”
“They really are. It’s the EMI. It’s the packaging of the sensor itself. It’s the single cable which can be robot mounted or static mounted. And it’s using the GigE signal connectivity instead of USB for the signals.”
“Got it. Thanks so much for your time today, Neil.”
“You are most welcome, Marianne.