In the past few episodes of Route to Reliability, we've talked about the potential for automation and robotics to economically access hard-to-reach places and record exponentially more data.
With that, some inspections are terrible applications for automation, forcing technicians to default to manual inspection.The scope of work may be too small. The space or access is such that your automated platforms literally won't fit. Or maybe there's a complex geometry, like a manway, that merits a deep inspection, with an extremely agile tool in order to take the right readings.
When we encounter these circumstances, how do we maximize the value of the manual deployment? How do we upgrade the technician’s abilities? It’s been up to them to not only record the measurements, but record the position, facilitating reproduction and repeatability.
HEBI Robotics offers tools for inspection technicians with the potential to supercharge the resulting data from manual inspection. HEBI has entered the NDT market with a passive articulated arm known as the Mobile Absolute Positioning System (MAPS). A technician in the field can attach a sensor to the business-end of the MAPS arm, and proceed with the inspection as required. This multi-jointed arm is able to not only log measurements, like say a UT A-scan, but also log the exact position of the measurement in 3D.
This allows inspectors to easily acquire the same amount of data density and positional accuracy, rivaling the capabilities of automated inspection systems.
Below are excerpts from my conversation with Dave Rollinson, HEBI Co-Founder and Bob Raida, COO.
Listen to the full episode!
Tell us how HEBI was formed and a little bit about your focus.
HEBI Robotics produces a platform for robotics development. The technology was developed at the Robotics Institute at Carnegie Mellon University, to make it easier to create customized robots, originally for snake robots. In fact, that's where the name “hebi” comes from. It's the Japanese word for snake.
Snakes can be really difficult to build, maintain and repair, so the team created a platform, allowing them to be created in a more modular fashion. The platform evolved and became applicable to other types of robots, not just snakes. We apply our robotics to NDT inspections and maintenance applications, where the high level of variability makes it difficult for conventional robots to be effective.
I'd really like to talk about the modularity aspects of what you guys are doing. This could be the foundation for a number of reliability solutions, but tell us a little bit more about the modularity. That could really accelerate development of robotic technologies and make the costs a little bit more manageable from an R&D standpoint.
That’s exactly right-- our tagline is “accelerating the future of robotics.” We came out of a world where we saw the gaps between buying an “off the shelf robot” and trying to do unique things with it.
The alternative is building everything from scratch. To do that, you have to be an expert in robotics and so many different areas just to get out of the gate. Allowing people to put things together intuitively and quickly gives them a better starting point to try things in the real world.
That's where reliability comes from. It comes from testing and iteration in the field, from the people that are using the technology.
Let’s say you have a six degree of freedom arm for example, and you need a seven degree of freedom arm, or a five degree of freedom arm. With a conventional robot, you've got to go back to the drawing board. You need to redesign it or find a different solution. With our system you can swap out joints, add joints, extend link lengths and it's just a couple lines of code.
R&D is an expensive hobby so it sounds like you guys are making it a little bit more manageable.
Did you set out from the start to be an industrial robotics company? Or were there other applications that you explored before you landed on this?
We started off thinking we were going to commercialize the snake robot itself for things like industrial inspection, search and rescue, pipes, power plants-- basically anywhere you can't easily fit a camera.
Getting into NDT was a happy accident. We spoke with someone in the industry and he described the pain points of inspecting complex geometries with relatively simple probes. He needed custom configurations to fit into weird nooks and crannies, because no two pipe elbows, for example, are the same in this field.
Think about manufacturing and automation inside a factory-- that robot is going to be stationary. It needs to do the same thing a million times, exactly the same way. That doesn't really speak to modularity or force control because that's more about rigid precision. In the real world, the applications you want to automate require more human-like motion. They're highly variable. Inspections and maintenance are perfect examples.
When I got my initial experience in NDT, it was doing manual readings on a lot of oddball asset components that needed to be monitored constantly. You're painfully aware that you're relying on the human operator to not just take the measurement but to accurately log that measurement, to allow for reproducibility.
Exactly. So the idea is, you go there and learn what needs to be done, which isn't just running the probe smoothly on the pipe. The job to be done is figuring out how thick the elbow is and where it’s gonna fail.
Having confidence at the end of the day boils down to one number, right? You gotta have confidence in that number and that's what’s lacking in a lot of places.
It's all about having confidence and the context around that number, both for the asset owner, but even for the person in the field.
I inspected sewers in a life before HEBI and returning home from the field, we didn't really know if the data was good. We’d think about how we might have to go out and inspect it all over again. It was nerve wracking.
At HEBI, we realized that we didn’t need to automate the process. We needed to provide a tool to augment the person and measure what they're doing in a way that wasn't currently possible.
That led to what we call MAPS, a modular, absolute positioning system. We stripped the motors and gears, leaving the encoders that measure position. It's a passive arm, so you are the one that drives it, but it provides a lot more flexibility. You can get around obstacles and weird surfaces.
This is not designed to replace people, it’s designed to augment the technician and make them superstars.
So I find it noteworthy that you're part of the Pittsburgh tech community. Do you have thoughts on developing technology in Pittsburgh?
There's definitely a mindset that rubs off on the people that live here. It’s one of being a lot more practical and realistic.
I'll try to be nice, but when you go out west to the other coast, there are really amazing things that happen there. But there’s a specific mindset of go big or go home and only pursue the super high risk, high reward problems. With that comes a little bit of a lack of respect for the power of finding a good niche and building up expertise. There’s also a lot of talk, a lot of show, a lot of pizzazz.
Pittsburgh is more of a heads down, do the work and keep it practical mindset. On the flip side, we should probably pop our heads up every now and then to talk about what we're doing, and put the word out a little bit more.
I couldn't agree more. There's a heavy blue collar component to it that Pittsburgh lends itself well. I can see other markets not necessarily being so dialed into that mindset and that buying process.
One of our killer advantages early on, was how everyone was like, “you have to manufacture overseas.” Without realizing that makes sense at a certain scale, a scale that's a million times larger than we are right now.
We decided to keep the vast majority of our manufacturing processes in house. The development, final assembly and iterations.
Again, that's the Pittsburgh mindset, “we'll just build it ourselves.”