Smarter Robots, Better Automation

BH: Welcome to another episode of Enabling Automation. I’m your host, Ben Hope. Today

we’re talking about robotics in assembly, but probably not in the way you might expect. Robotic capability has advanced tremendously through technologies like vision, sensing, collaborative, and software. Yet developing automated assembly systems still requires significant engineering effort. Today’s conversation isn’t simply about robots. It’s about innovation in what robots can do and innovation in how we build and deploy automation systems. Today’s guest is Nathan Snider. Welcome to the podcast, Nathan. Please take a few minutes and introduce yourself.

NS: Thank you very much. I’m Nathan, I’m an applications engineer here at ATS, been with the company over five years now. I really work on a lot of custom engineered to order life science opportunities, doing a lot of different concepts for high speed medical device assembly, primarily. But really we get into lab automation, consumer products, all sorts of different applications. So every day is different and always excited to work on automation projects and learn about new trends in the industry.

BH: Excellent. Okay. Well thanks for joining us today.

NS: Looking forward to the conversation.

BH: Okay. So before we discuss new technologies, let’s set the stage. Why is assembly still one of the hardest things to automate?

NS: I think with assembly there’s a lot of different challenges. So, One. Products are always changing and customers are coming with new requests. They want more features. Doesn’t seem to get easier, which is a good thing for us because it makes our job interesting and gives us new challenges to solve. But really, it’s the assembly. Automation has been around for a long time, but you’re challenging the technology with higher precision, higher speeds, things that maybe haven’t been automated before, or they were expensive or not very feasible to automate in the past that you’re now with new technologies, trying to take that next step and automate things for the first time.

BH: Figure it out. What can a person do instinctively in an assembly process that is still maybe difficult for a robot to do?

NS: Really like the feedback I guess you get when you pick something up with your hand and you’re able to see, oh, that clips together or something didn’t work out the way you want to be able to adapt quickly. If you’re looking at that in an automation sense, you can definitely incorporate sensors and feedback and vision systems to be able to do all those things that people intuitively do. But it’s a lot simpler to just pick something up, sometimes put it together versus trying to engineer everything out of the solution.

BH: And all the software and programing around that to do that and manage it.

NS: And yeah, yeah, for sure. There’s, there’s a lot of engineering effort.

BH: Are there assembly applications that maybe look simple but become surprisingly difficult once you kind of start getting into it?

NS: Yeah, definitely. One thing I can think of is just flexible components. It might look simple, especially if you’re looking at it on a CAD drawing. You’re just going to get lots of clearance. You can pick something up, load it into a different position, but when you actually start playing with the parts and seeing how they handle, that’s when things can become a lot more challenging. And we definitely have a lot of experience working with flexible feeding solutions and all kinds of different solutions to get things assembled properly, but can definitely overlook that, especially in the front end if you don’t have those parts to play with.

BH: Yeah, variability can be a huge challenge. I think that’s really the horizon of the next stage of automation is how do you deal with all the variability and whatnot?

BH: Where does the complexity usually come from? The robot? The process? The parts? Or the system around it?

NS: Probably a little bit of both, but definitely the products that customers are coming to us with are more challenging. And some of the processes haven’t been automated before. So again, this is sometimes where you have a process that could have been done for many years, but automating it is only now becoming more feasible just from a return on investment point of view. We have better technologies that we can work with. So really that would be kind of my main take on that.

BH: Okay. So we’ve established why assembly is difficult. Let’s talk about what’s changed. So what can we automate today that we maybe couldn’t have done ten years ago.

NS: Going back to the flexible components, as some flex feeding technologies have come a long way. Vision and AI feedback, different technologies there are definitely a lot more usable today than they might have been ten years ago.

BH: You don’t need a PhD to program a vision system anymore.

NS: Yeah. Yeah, exactly.

BH: Is the robot still the most innovative part?

NS: I’d say robot innovation is definitely part of automation, but robot is just one piece of the puzzle and the platform and the software and everything else behind it. I would say it’s probably more important today than it was in the past. The robots, there’s definitely been innovations. They’re getting faster, more accurate, easier to program. But in terms of what they’re capable of for a lot of applications, that hasn’t changed significantly over the last ten years.

BH: Yeah, yeah. They need to understand how to move like the innovation got into the motion of the robot is maybe not as innovative as it once was, but now it’s all the different technologies that are feeding how it decides where to move, how to move, etc., etc..

NS: And with all these new sensors and having more feedback, the robot becomes a better tool because you can give it more feedback and it is a flexible, servo controlled system that can adapt. But again, if it doesn’t have those sensors in that input, it doesn’t necessarily know what what to do.

BH: It’s just a dumb device at that point. Yeah. Yeah. So that’s interesting. So it kind of back maybe ten years ago, the big kind of focus was repeatability. But today maybe the focus is adaptability for that variability in parts and things like that.

NS: Yeah. Robots are definitely getting very repeatable. And maybe there’s some applications and they definitely are where there’s still challenges there. But with incorporation of vision and being able to adapt, if you have a repeatable robot, you’re usually pretty good for the majority of your applications.

BH: And it seems like people are maybe more up for the challenge than they used to be in the past. They abandoned an idea if it was going to be too challenging. But now with the technology that’s available, people like we could probably figure it out. Let’s do a proof of principle. Let’s do some testing and experimentation and things.

NS: Yeah, and I’d say we thrive on that kind of stuff at ATS. Like really, that’s what gets me up in the morning coming to work.

BH: That’s the fun part.

NS: Yeah, yeah. So it might be something you want to turn away from initially, but if you spend a bit of time on it, you’d be surprised what you can come up with.

BH: Yeah. Yeah, that’s very cool. So what do you think is a better approach kind of going forward engineering variability out or teaching robots to adapt? And maybe that’s a pretty easy question.

NS: Yeah I think a bit of both. Like obviously if you can engineer your end-of-arm tooling and the things that are interfacing with the product to account for larger variations and components, that’s always going to help. But if you have a system that can take feedback, adapt and then recover when something goes wrong, that’s very important as well.

BH: It opens up immense possibility. Okay. Collaborative robotics. And I think that’s a topic that changed the conversation. What do you think was the biggest innovation when it came to collaborative robotics?

NS: So I think working alongside robots was definitely there was a lot of hype initially. You can work with a robot and not have to have guarding, but I think as people started to use them, it’s really having something that’s easy to program and you can set up without necessarily needing to be a super senior programmer. That’s one of the benefits I’ve heard from the people on the floor. But ultimately, collaborative robots aren’t maybe being used as much in those right beside operator scenarios as we might have thought initially. You see it a lot more so in the packaging industry, but when you need to go faster then sometimes, it makes more sense just to have the system guarded.

BH: Yeah, it’s interesting because I think like ten years ago I was at Hanover Messe and the big topic was collaborative robots and everything. Everybody was talking, but the huge emphasis was on the interface and being able to get to the point where maybe even the operator could program the robot themselves and change things themselves. And yeah, to your point, I think that showed people how easy it could be and then where we could take that kind of usability and apply it to other areas, because you want to get automation running as quick as possible and as straightforward and easily as possible. You don’t want to deal with complexity when you don’t need to.

NS: For sure.

BH: That’s really interesting.

NS: Yeah, if you have a robot, whether it’s collaborative or not, if it allows people to program it quickly and efficiently better than they would with another model, then there’s always going to be a benefit to that in my mind.

BH: Absolutely. It’s an intuitive kind of application. And then you really only need to understand your process. You don’t need to understand all the technologies that kind of relate to the process. So that’s cool. So it’s easier deployment than the real innovation. Would you agree with that?

NS: Yeah, I think so. For sure. Having a robot that’s easy to implement is always going to be a benefit. Having a robot that you can physically touch safely is also a good thing to have in certain applications, but it just isn’t something that you always need, and sometimes it’s more prohibitive due to the speed.

BH: Exactly, exactly. And then. Yeah, and I think productivity is you want to be productive. So I think collaborative robots are cool. Maybe when you’re doing integration and there is people around, you could kind of have everything slowed down and still kind of work. Once you get to production you want high speed and etc. etc.. So better robots do not mean better or easier automation. Where does engineering time go when you’re trying to commission and get an automation system up and running?

NS: I’d say a lot of your time up running when you’re engineering a system is really being aligned with the customer on the process and all of their needs. We could do things faster right out of the gate if that’s the end goal. But if there’s things that are missed along the way that aren’t communicated, then the end product isn’t necessarily going to be accepted by the customer or they won’t. They’ll be missing things that they really wish they would have put in. So as much as we we can go fast and want to go fast, it’s a big part of it is just listening and understanding and making sure everything’s captured correctly.

BH: User requirements are tight, everybody’s agreed on them and yeah, yeah, yeah, okay.

NS: You might have somebody who puts together a URS (User Requirement Specification) in matter of a few days because they’re on a rush project, and then that’s the basis for the quotation. And then a larger engineering team gets involved from the customer standpoint. And there’s things that may have been interpreted differently. So having that communication and alignment is really important. And we definitely do a good job of that in the front end. But kicking off a project you want to make sure that everybody is for sure aligned.

BH: Is happy. Yeah, it’s an expensive ordeal. And I think everybody obviously wants the right machine at the end of it.

NS: For sure.

BH: How much in the overall kind of cost of an automation system is contributed to kind of robotics and tooling?

NS: That’s a tough question because I think it kind of depends. Every everything’s different. It might be half of your cost in some applications and it might be 10% in others. If you have a lot of complex engineering where you’re going to be spending weeks and weeks designing and integrating something, then obviously that’ll skew your cost a little bit more towards the engineering side than the material and tooling cost of the robots.

BH: Right, right. It all stems from the application and the requirements from the application.

NS: Yeah.

BH: What do you think the biggest integration challenges are with getting automation going?

NS: So I think going back to things like flexible parts, more complicated customer devices that are being automated for the first time. So you might have tiny little components. And sometimes the actual tolerances of the components don’t match the drawings. So you’re working through those kinks in the integration phase for the first time and maybe needing to make some adaptations. But really, just having a complicated products is, I think, one of the challenges that we face.

BH: So why aren’t systems easier to build then? Why do you think?

BH: I would say in some senses it’s easier because we do a good job at standardizing on certain platforms and standardizing with our code, allowing us to design and integrate quicker. But at the same time, we have all those tools at our disposal, but then you’re met with new requirements, new processes, smaller components, and some of these things you’re integrating for the first time. So that’s why, although it may not really be a whole lot different in schedule, if we were to have tried to do something that we’re doing today ten years ago, it likely would have taken us longer because we don’t have all those tools at our disposal.

BH: Right? And sometimes just the physics of the real world can kind of give you an unexpected slap in the face that you maybe don’t always consider.

NS: For sure. Yeah, our designers are good, but they can’t anticipate everything. And that’s just part of the automation world.

BH: Yeah, that’s the journey. Yeah. Okay. You mentioned kind of platforms and platform adoption. So innovation really in my opinion also means improving how we deploy automation. So do you think there is an opportunity to improve how we deploy automation?

NS: Yeah, I think really just standardizing with your designs and your platforms can allow you to be more efficient, especially if you get an application that you’ve done something very similar to. You’re in a very good position to execute that very quickly, more cost effectively, getting the customer a stellar solution faster.

BH: Yeah. Reuse assets, reuse designs. And yeah, when you get to integration then you’re testing things. Or maybe you’re not testing as much as you need to because you’re using validated code, validated technologies to kind of get where you need to go.

NS: And I think platforms play a big part in that. Like we talk about SuperTrak that we have here where we have different versions for different applications and it’s very versatile. You have Symphoni for high speed automation platforms and just utilizing standard products that we’ve implemented many times before, whether that be a specific robot or other pieces of third party equipment. Those are all things that are going to help us be more efficient.

BH: Yeah. And even like right from conceptualization, quoting faster, being more confident going into the design phase that you have the right approach and you can kind of move forward.  So what do you think could be standardized in kind of automation systems? What kinds of technologies do you think are good candidates for standardization?

NS: Definitely your conveyance platform is a good one. Assembly robots, whether that be custom designed pick and place that you reuse multiple times like a Symphoni RSM (Rapid Speed Match) arm, or a third party robot.

BH: Just have different strokes and kind of payload kind of capabilities.

NS: Yeah, yeah. And then I’m really interested as we’re moving forward, we have a lot of standard code and different libraries that we pull from and designs that we pull from. But with AI becoming more apparent, using AI to help us implement faster and really just make us more efficient.

BH: Yeah. Very cool. Okay. And then like I always find like you can standardize on the kind of small little bits and pieces like air prep, valve terminals or mode IO (Input/Output). You’re not spending time trying to select this. You can just have different kind of options for IO or whatever. Do you think you can standardize how you solve applications?

NS: I think so. I would like to say we have a pretty good handle on having a process, whether it’s in the front end, in engineering or integration, where you follow a process that’s proven itself over many years, and you’re able to solve problems and adapt to things that come up.

BH: Why do you think platform adoption isn’t more common?

NS: Sometimes you want to use a certain platform because it’s what you know, but occasionally you’ll run into a situation when it’s really not the best fit, but you’re still moving in that direction. So you still need to be open minded and flexible, even though you have all these tools and platforms that are great in a lot of scenarios. Sometimes you need to just have a clean set of constraints and think about what the solution could be and get on a whiteboard and figure it out.

BH: So let’s finish the conversation by looking to the future. And when you see automation, what kind of developments or kind of innovations are exciting for you at the moment?

NS: Yeah, I think I mentioned AI. I know I’m starting to use it a lot more in my day to day, and I’m seeing the benefits we have where things that you used to be very manual, repetitive tasks, whether it’s in the front end when you’re responding to customer requirements or putting together designs or code. Definitely the AI tools are becoming a big benefit and I don’t see that slowing down.

BH: So. Yeah, saving a ton of time just to do those repetitive, mundane tasks.

NS: Yeah, 100%. For sure. So that’s definitely exciting. But I think, two, just you go down to the Innovation Center and you see all the new things we’re working on, whether it be high speed applications, new platforms we’re developing, or even just testing out a cool product that’s on the market; that’s never going to stop. I think there’s always going to be cool things, and we’re always willing to try them out. And sometimes it’s just on our own. We don’t have customers coming to us, but we see the value in some of these technologies and getting ahead of it. So we’re in a position where we can use it when the right opportunity comes along. That’s really cool.

BH: Get the mindshare, get the technology into people’s minds so they understand how to utilize it. And then, yeah, when the opportunity comes you go with it. That’s cool. Do you think there are tasks that humans can do but robots still really, really struggle with?

NS: I would say there’s definitely still some applications where it’s not that a robot can’t do it, but the feasibility and all the sensor feedback that would be required, the time to implement it just doesn’t make a lot of sense in certain cases. So yeah, it’s not it’s not always a slam dunk.

BH: Yeah, yeah. We’ve worked with customers that are doing things with like cable harnesses at different lengths, and they found that they needed these robots with huge reach and huge payloads, even though the payloads nothing, just to be able to manage the length of the cable harnesses. And often it’s just easier to do it manually, just get a human to do it rather than spend like $100,000 on this huge robot to get the reach that you need.

NS: And then there’s that side of it. It’s going to take a long time, but there’s also a risk that you put this thing together and it may or may not work as well as you anticipated, where something…

BH: And you’ve spent a lot of money on it.

NS: Yeah, yeah, yeah. Where if you just do that process manually obviously that has pros and cons, but occasionally that is the right approach.

BH: So kind of looking forward,  do you, do you see kind of technology moving towards smarter robots or better platforms to build kind of automation on top of, or maybe both?

NS: Yeah, I hope both. It gives me more tools to present to customers. And if we can have products that are more adaptable and can adapt to all this change and higher speed, higher precision with feedback and fault recovery, that’s great. And then having a solid platform that you can implement these things on time and time again to become more and more efficient and give the customer a tried and true solution, then that’s definitely exciting.

BH: Yeah, like the usability I find, because younger generations expect things to be intuitive and to work. And in my career there’s been systems I’ve worked on which has so many cryptic faults. And like, you need a PhD just to understand what the fault is and how to recover from it. And now I think the expectation is the machine should tell me what happened, machine should tell me how to fix it, or just fix itself and recover. I think in my experience, like getting something to work is step one keeping it working and getting a recovery that’s kind of intuitive is the biggest challenge.

NS: For sure.

BH: And keeping that going. And then when you use platforms, a lot of that work can be done for you and then you can move forward and focus on the parts that require the customization, the end-of-arm tooling and things like that.

NS: And those types of systems not only benefit us when we’re building the equipment, but when we give it to our customer, makes their lives a lot easier and they’re much more happy with the end product. So we’re definitely excited to try to improve.

BH: Yeah, I’ve talked to lots of customers and they struggle with turnover on operators. And like you spend six weeks training somebody and then they leave and then you got to start over again. So having machines where you don’t really need the same amount of training, you can be like to operate this machine, you press this button. When this happens, you do this, you do that. And that goes a long way for many factors. Interesting. So Nathan, thank you very much for joining us today. One of the biggest takeaways for me is that robotics continue to evolve on two fronts smarter robotic technology and better ways to engineer automation through standardization, through reusing assets, and through platform thinking. Ultimately, smarter robots are only part of the solution. The future of assembly automation will also depend on how effectively we deploy those technologies. Thanks for listening to Enabling Automation. If you’ve enjoyed today’s conversation, subscribe and join us next time. Thank you.