Episode Show Notes
Episode 1 features Dr. Satyandra K. Gupta, Smith International Professor in the Aerospace and Mechanical Engineering Department and Director of the Center for Advanced Manufacturing in the Viterbi School of Engineering at the University of Southern California; and Craig Schlenoff, Group Leader of the Cognition and Collaboration Systems Group and Program Manager of the Robotic Systems for Smart Manufacturing Program in the Intelligent Systems Division at the National Institute of Standards and Technology (NIST). SK and Craig discuss the evolution of robotics technologies, the impact on the manufacturing industry, how to develop a roadmap for implementation, and their predictions for the future of robotics.
Dr. Satyandra K. Gupta is Smith International Professor in the Aerospace and Mechanical Engineering Department and the Director of the Center for Advanced Manufacturing in the Viterbi School of Engineering at the University of Southern California. Prior to joining the University of Southern California, he was a Professor in the Department of Mechanical Engineering and the Institute for Systems Research at the University of Maryland. He was the founding director of the Maryland Robotics Center and the Advanced Manufacturing Laboratory at the University of Maryland. He also served as a Program Director for the National Robotics Initiative at the National Science Foundation from September 2012 to September 2014. Prior to joining the University of Maryland, he was a Research Scientist in the Robotics Institute at Carnegie Mellon University.
Craig Schlenoff is the Group Leader of the Cognition and Collaboration Systems Group and the Program Manager of the Measurement Science for Manufacturing Robotics Program in the Intelligent Systems Division at the National Institute of Standards and Technology (NIST). He leads the Agility Performance of Robotic Systems project and co-leads the Embodied AI and Data Generation for Manufacturing Robotics project. His research interests include knowledge representation/ontologies, intention recognition, and performance evaluation techniques applied to manufacturing robotic systems. He has led multiple million-dollar projects, dealing with performance evaluation of advanced military technologies and agility performance of manufacturing robotic systems. He has published over 150 journal and conference papers, guest-edited three journals, and written three book chapters. He is currently the Associate Vice President for Standardization within the IEEE Robotics and Automation Society and the vice chair of the IEEE Robot Task Representation Working Group. He previously served as the program manager for the Process Engineering Program at NIST and the Director of Ontologies at VerticalNet.
Highlights
00:00:00 - Introductions
00:02:25 - When robotics technology started becoming integrated into manufacturing processes
00:03:19 - Expected impact of robotics on businesses and society in general
00:06:27 - Realities of robotics capabilities
00:10:22 - Examples of robotic technologies or initiatives that can have a positive ROI today for the smaller manufacturer
00:11:09 - What makes those applications able to achieve a positive ROI
00:12:46 - Different types of robotic arms
00:13:40 - How to identify if robotics is right for a given production environment
00:15:53 - Definition of cobots and benefits versus traditional robots
00:17:36 - What makes the collaboration possible
00:19:39 - What improvements in cost, quality, and productivity you would expect to see from robotics
00:23:16 - How to develop a roadmap to identify and implement robotic solutions in your manufacturing process
00:27:56 - How long it takes to program a robot and skills required
00:30:32 - Reliability and maintenance
00:33:41 - Ensuring that humans are safe working around cobots and robots
00:37:57 - Future of robotics in manufacturing sector during next decade
Transcript
Gregg Profozich [00:00:00] In the world of manufacturing, change is the only constant. How are small and medium-sized manufacturers, SMMs, to keep up with new technologies, regulations, and other important shifts let alone leverage them to become leaders in their industries? Shifting Gears, a podcast from CMTC, highlights leaders from the modern world of manufacturing from SMMs to consultants to industry experts. Each quarter we go deep into topics pertinent to both operating a manufacturing firm and the industry as a whole. Join us to hear about manufacturing sectors' latest trends, groundbreaking technologies, and expert insights to help SMMs in California set themselves apart in this exciting modern world of innovation and change. I'm Gregg Profozich, Director of Advanced Manufacturing Technologies at CMTC. I'd like to welcome you.
Gregg Profozich [00:00:48] In this episode, I'm joined by Dr. Satyandra K. Gupta, Smith International Professor in the Aerospace and Mechanical Engineering Department in the Viterbi School of Engineering at the University of Southern California, where he is also the Director of the Center for Advanced Manufacturing. I’m also joined by Craig Schlenoff, Group Leader of the Cognition and Collaboration Systems Group and the Program Manager of the Measurement Science for Manufacturing Robotics Program in the Intelligent Systems Division at the National Institute of Standards and Technology (NIST). SK and Craig discuss the evolution and impact of robotics technologies, how to develop a roadmap for implementation, and their predictions for the future of robotics.
Gregg Profozich [00:01:30] Welcome, Craig. It's great to have you here.
Craig Schlenoff [00:01:32] It's a pleasure to be here.
Gregg Profozich [00:01:33] Craig, can you take a minute or two and tell us a little bit about your current role?
Craig Schlenoff [00:01:36] Sure. I am a program manager at the National Institute of Standards and Technology, which is part of the Department of Commerce in the US. My role is I lead the program focusing on manufacturing robotics. We have about a seven-and-a-half-million-dollar program, looking at different aspects of manufacturing robotics — anything from agility to perception to sensing to many other areas. I'm also a first-level manager, which is a group leader at NIST, focusing on collaboration types of systems.
Gregg Profozich [00:02:04] Thank you so much. It's really great to have you here today. Welcome, SK. Glad to have you here, as well.
Dr. SK Gupta [00:02:08] Truly a pleasure. I'm very happy to participate.
Gregg Profozich [00:02:11] SK, could you take a minute or two and tell us a little bit about yourself and your current role.
Dr. SK Gupta [00:02:15] I'm a professor in Viterbi School of Engineering within the Department of Computer Science and Mechanical Engineering, and I also direct Center for Advanced Manufacturing at USC.
Gregg Profozich [00:02:25] Thank you so much for being here. Really love the fact that your backgrounds are going to lend perfectly to what we're trying to talk about today. Looking forward to our conversation. Let's get started. We're here to talk a little bit about the evolution of robotics technologies in manufacturing. Let's go back to the beginning. When did robotics technology start becoming integrated into manufacturing processes? SK, can you start us off?
Dr. SK Gupta [00:02:45] Sure. One of the earliest examples of robot deployment was a Unimate robot, which was deployed by General Motors in early '60s, and it was used in material handling application. From there, robotics grew in many more applications.
Gregg Profozich [00:03:01] It's been around 60 years and counting from its infancy to the current state. We'll talk about that in a few moments. Craig, anything you'd like to add along those lines?
Craig Schlenoff [00:03:09] There's been, I think, pretty dramatic growth in robotics over the last decade or so. It's been around for quite some time, but I think this is about the timeframe where we're really seeing a lot of the benefit that is provided.
Gregg Profozich [00:03:19] Let's talk a little bit about some of the current trends as robotics technologies evolve — like you said, the last 10 years has been a very quick part of the evolutionary period — and they're adopted more broadly. What impact do you expect them to have on businesses and on society in general?
Craig Schlenoff [00:03:34] Robotics are not going to be able to do everything. There's certain things that they're good at. But one of the things that I think would be most impactful, I think, for society in general, is that they're likely going to take away some of the more mundane jobs that people tend to do. To give you a story, about 5 or 10 years ago, I went to a factory, and in the factory there was a lady. The lady's job was to pick up a part from one table, and take a couple steps, and move it over to another table. She did that thousands of times per day. That was her job. Those are the types of jobs that robotics are very good at, and those are the kinds of jobs that typically, people don't want to do. In speaking with the person that was running the factory, they mentioned that they had a very high turnover in that type of position, because it's a boring job, for lack of a better way of saying it. There's a lot of repetitive motion and things that people don't want to do. That's just one example of how I think it's going to impact businesses and society is that it's going to take jobs that often people don't want to do that are repetitive and mundane, and we're going to be able to automate them to free them up to do more interesting jobs.
Gregg Profozich [00:04:36] Have the robots do the dirty, dull, and dangerous, the 3Ds trifecta for robotics, and really free humans up for more creative things that affirm our dignity, if you will.
Craig Schlenoff [00:04:46] Exactly. You mentioned dangerous. Dangerous is a big part of that, as well. There's a lot of jobs dealing with chemicals, and gases, and things of that sort that are not safe for people to do. For those types of jobs. robotics, I think, will play a big role in not only doing things safely, but also, as you said, freeing up people to do more important jobs or maybe different jobs.
Gregg Profozich [00:05:03] SK, how would you add to that?
Dr. SK Gupta [00:05:05] Well, three different other dimensions to it. One is that we get significantly superior quality as a result of robot being consistent. Robot does the same thing over and over, and therefore, you get very consistent performance. That then leads to superior quality. The second challenge that we seem to be seeing nowadays is fluctuations in demand. When demand fluctuates, staffing very quickly with human workers and then letting them go when your demand goes the other direction is challenging, while if you have robots, it gives you significant capability to handle demand fluctuations. If you have a robot, you can start a second shift with no problem; third shift, no problem. It gives you inherent capabilities of dealing with demand fluctuation. Finally, robots can do things which human cannot. It gives you the ability to transcend human abilities. You can work in extreme cold environment; you can manipulate very small things; you can manipulate very large things. Once you can transcend human abilities, then that naturally leads to innovation. I would add to making workplace safer, which Craig pointed out, three other things, which is significant improvement of quality, ability to handle demand fluctuations, and enabling innovation.
Gregg Profozich [00:06:27] Connected to that, in my mind, at least, there seems to be a perception in society that robots are going to take away people's jobs. In some cases, like the tasks you mentioned, repetitive motion injuries, those tasks should be replaced but not necessarily the jobs. How do you put a cost on human suffering is the issue there. My understanding is that robots really can't replace us yet. Let's talk for a few minutes about the realities of robotics capabilities. What can and can't they do?
Craig Schlenoff [00:06:53] Let me start by saying that robots will never, in my opinion, completely replace humans. You said a little while ago that robots haven't replaced humans yet. I don't believe they're ever going to replace humans. There's things that humans can do that I don't believe that robots ever will be able to, things such as deep understanding and reasoning about the environment, and be able to take steps based upon new situations that they're confronted with. Robots are taking steps in that direction, but I don't think we're ever going to get to the point where they're going to replace humans. What they are good at now are things, as I mentioned before, very repetitive tasks with low variability. That's really the place that we want to start. The robots are performing those types of operations now. They're doing it primarily in the areas of things such as packaging, and palletizing, and material handling, and welding, and assembly. They're starting to get to the point where they can handle variability, variability in the sense that parts are in slightly different locations than they were expected to be or you need to put things together a little bit differently. They're starting to move in that direction but nowhere close to the point of being comparable to what a human can do. In summary, there will always be a need for humans. But I think robots are starting to get to the point where they can be a little bit more creative and a little bit more intelligent. But still, their strength now is in the repeatable low variability tests.
Gregg Profozich [00:08:07] SK, what do you have to add to that?
Dr. SK Gupta [00:08:09] Robots are very good in terms of being able to move things. But to require tasks that require dexterity, where it becomes extremely important to interact with the objects in a controlled manner, now that brings them immediately ability to sense is based upon that sensing react. Now, if you think about taking a soft object or some compliant object and you start manipulating it, you cannot preprogram emotion that a robot can follow. Robot will start interacting with it, feel it, sense it. Since robots' ability to think and sense are limited. Robot have excellent ability to move. They can apply a lot of force. They can move at a very high speed. They're very accurate. However, their ability to sense and their ability to sync is very limited. That means that any task that requires in addition to motion, thinking part and sensing part, that's where robots are currently not very good at.
Gregg Profozich [00:09:08] I hear you guys say that something like picking a part out of a bin can be very difficult for a robot. A human three-year-old could walk up and grab just one piece, but a robot can't go in and grab a piece unless it can sense and figure out which way the piece is oriented and figure out how to orient the gripper or the end effector that has to go in and grab it. It's those kinds of things that are very complex to program a robot to do — sense, respond; sense, respond; adapt and adjust — on a regular basis that we do quite naturally. Is that what I hear you saying?
Craig Schlenoff [00:09:37] That's absolutely right. You're absolutely right.
Gregg Profozich [00:09:38] Craig, I now understand better why you say they'll never take our jobs, because it's a lot of computing power, and how do you plan for every eventuality of what a robot could run into? Put a robot on a moving cart, and move it around the shop. It still has to sense its environment, and it has to adapt to every change. We're built for that adaptation; they're not.
Craig Schlenoff [00:09:57] It's really amazing to think — it's just like what you said — what goes into a human's brain and what they think about when they have to do what we consider to be very simple tasks: picking something thing up, knowing how heavy it is, knowing how far the human can reach, knowing where they need to put it, knowing where they need to grasp it. Things that we do naturally, a two-year-old does naturally, is very, very complicated for a robot to do because of all the knowledge that needs to go into that in order for it to do it.
Gregg Profozich [00:10:22] Well, I feel better now. My job is safe. Let's talk a little about implementations. I know a lot of large companies have implemented robots to a positive ROI. What are some examples of robotic technologies or initiatives that can have a positive ROI today for the smaller manufacturer?
Dr. SK Gupta [00:10:39] There are parts such as material handling, machine tending, inspection where you can deploy a robot and where users themselves can program the robot. You don't need to hire a system integrator to program the robot. There are tasks which users can themselves program the robot. Therefore, you can see immediate benefit of deploying the robot and see positive ROI.
Gregg Profozich [00:11:03] Material handling, inspection, palletizing, I'm assuming, machine tending, those kinds of things?
Dr. SK Gupta [00:11:09] Yeah.
Gregg Profozich [00:11:09] What makes those applications able to get to a positive ROI? Is it the state of the technology? Is it the ease and consistency of the job that's being done?
Dr. SK Gupta [00:11:19] Usually, many of these tasks just require a single robot. You're not trying to coordinate among lots of different machines. Your motion accuracy need not be super precise. When you come close enough to the part when the gripper closes or suction gripper positions over the part, you can still grab the part. You have quite a bit of tolerance in terms of what level of accuracy you need from the robot. Therefore, programming becomes much, much easier. Also, there's a pattern. Even though you may be processing different parts from day to day, but the motion patterns remain very, very similar in nature. Your cells are simple. You're very tolerant to the accuracy that you need in your motion. Generally, the motion patterns that robot meets are very similar. Therefore, programming becomes much, much easier, and system integration is not a big challenge. Those are the commonalities shared by those applications.
Gregg Profozich [00:12:16] We're looking for those places where there's a similar motion path and a similar level of accuracy or error allowed for the end effector to be able to grab and manipulate the part.
Dr. SK Gupta [00:12:26] That's correct.
Gregg Profozich [00:12:27] To go off on just a bit of a tangent, we're starting to use some terms here I want to make sure everybody's comfortable with. If I was a small manufacturer and went out and bought a robot, I would buy just the arm, correct?
Dr. SK Gupta [00:12:36] You will buy an arm. It's necessary. You will buy a small PLC and a small camera if you're going to be doing machine tending.
Gregg Profozich [00:12:36] For machine tending, likely a camera. But then there's the robot manufacturers out there in the world, the ABBs, the FANUCs, the Yaskawa, the Productive Robots, the Universal Robots. They produce the arm, but they don't put the hand on the end of it. We talk about the end effector. What are some of the different types that can be applied to a given robotic arm?
Craig Schlenoff [00:13:03] There's standard types of robot grippers. There's vacuum grippers to work on a vacuum where they go, and vacuum goes into effect, and it sucks up the part and attaches to it. There's pinch grippers or two-finger grippers, where they can go with just one finger on each side and pick up the part. Those are more of the standard types of grippers. But sometimes, depending upon the application, you need very specialized grippers, grippers that are made specifically for the part that you're trying to pick up depending upon how you want to pick it up. Those get to be a little bit more expensive, a little bit more time-consuming in order to produce, but very customized towards your type of part, depending upon what type of accuracy you need, similar to what SK was talking about earlier.
Gregg Profozich [00:13:40] How would you guys think about knowing if you were a small manufacturer if robotics is right for a given production environment?
Craig Schlenoff [00:13:47] I think the person will need to step back and look at the types of activities they're performing within their factory or within their shop floor. There's likely different types of activities. Going back to what I mentioned before, the repetitive tasks, low variability is where you want to start. It doesn't mean necessarily where you need to finish but where are you want to start. If you look at how robots have been used in large manufacturers, they're typically types of activities that the robot does tens, or hundreds, or thousands, or tens of thousands of times over and over again, like auto manufacturers and things of that sort. I'm not suggesting that a small manufacturer is going to create tens of thousands of objects. Maybe they are, but maybe they won't. But the important part, as SK was talking about before, is to look at things that you do over and over again, at least at some level of abstraction. Just like the discussion before about picking up different types of parts and moving it over to a pallet or things of that sort, even if they're different objects, it's the same type of activity. Those are things that would lend well to robotics. To start with the low-hanging fruit, if there are large lot sizes, even better, because it does take some time to program the robot. You want to make sure that the time that you put into programming the robot offsets the time benefit that you get by using it for the large lot sizes. Then work your way up. Then start looking at when maybe you don't do the exact same thing over and over again, but there's just slight variability. Maybe there's opportunities to apply robots to that, as well. One of the other things I want to mention is even if you don't have a large lot size, even if it is a small lot size, if it's something that you believe that demand will increase for your product, that might be another good opportunity for robots. If it's something that you're only selling a small amount of something but you think that demand could increase by programming that robot once, by having it do the right thing, you're able to increase your lot size with increased demand relatively easily, because you already have the infrastructure in place.
Gregg Profozich [00:15:30] The lot size is a big portion of this potentially, because it really talks about the volume it'd be doing and the repeatability. I think implicit in that is the idea that robots require setups between different kinds of operations. If I set it up once and run it longer, I have a better chance of getting productivity out of it is the concept here?
Craig Schlenoff [00:15:46] In general, yes. Some of the robots are getting quicker and quicker to set up, but it still takes time. You want to make sure you account for that time.
Gregg Profozich [00:15:53] In recent years in the world of robotics we've had a new term come into play — the cobot, the collaborative robot. Tell us a little bit about what collaborative robots are, and what impact they're having, and what some of the benefits of cobots versus traditional robots might be.
Dr. SK Gupta [00:16:07] Traditional robots, you have to put them in cages to make sure that humans are physically separated from them. Therefore, there was no chance of robot accidentally causing injury to humans. That meant that on the shop floor you need a lot of space to put the robots around; you needed to put appropriate controls in place to make sure that humans and robots can always be separated. That limited the applications where you could deploy it upon, because there are many tasks on which you cannot completely automate it. Therefore, a portion of the tasks need to be done by a human; a portion of the tasks need to be done by a robot. It would be very useful if human and robot can work on different facets of the task or different subtasks in close physical proximity. That can open up new sets of tasks where robots can be deployed. Cobots, or collaborative robots, they enable humans and robots to work together in close proximity. These robots, by design, ensure that if a collision were to happen... First of all, you try to prevent collision activity between humans and robots. Even if a collision were to happen, then the injury to humans will be very limited. Therefore, this opens up the possibility of humans and robots working in close proximity. Therefore, it expands the speed of tasks where robots can be used.
Gregg Profozich [00:17:36] What's the difference that makes the collaboration possible? What's the safety factor that comes in?
Craig Schlenoff [00:17:39] There's the standards that are out there. There are standards organizations that have published standards to say, what does it take to make a robot a cobot? How do you know that it's safe to work around humans? The standards, there's one in ISO, which is 10218. There's one in ANSI and RIA, which is R1506. They focus on a number of things. But the two main things that they focus on are something called force limiting and speed and separation monitoring. Force limiting is that if a robot gets too close to a human, it can put a certain amount of force on that human. There's all kinds of measures for what force it takes to cause pain within a human, or injury, or things of that sort. Make sure that if the robot's too close to the human, it can't exert over a certain amount of force. Speed and separation monitoring says that if a robot is too close to a human, it's got to slow down. Depending upon how far it is from the human, they would slow down so that if that happens where the robot accidentally bumps into the human, it's going slow enough so that it doesn't hurt the human. In order for these manufacturers that are developing the robots to be able to say that they're collaborative robots, they have to show that they conform with these two standards. Now, that sometimes introduces some challenges, one of which is some of the robots that need to pick up very, very heavy things like an automotive manufacturer or something, it's almost impossible for them to be cobots, because they need such force in order to lift these heavy objects that they can't really control that amount of force just to be able to pick up the object. They wouldn't be considered cobots. But there are many, many cobots that are out there that are usually able to lift lighter weights on the order of a few pounds that can conform to these specifications and therefore, are able to call themselves cobots.
Gregg Profozich [00:19:19] The force limiting and the speed and separation monitoring is the improvement that allows for the safety there, but there's a trade-off. Because it is moving slower and it has less force, it can't pick up heavier objects or move objects at a velocity that might be necessary. It will really depend on the application to decide cobot versus robot
Craig Schlenoff [00:19:37] Yes. Safety issues in the application, absolutely.
Gregg Profozich [00:19:39] Let's switch gears just a little bit. From a business standpoint, what improvements in cost, quality, and productivity would you expect to see from robotics? I would imagine that this would vary based on the robotic application, palletizing versus machine tending type of thing. But what are some of the general categories of cost improvements, quality improvements, productivity improvements that we might see?
Dr. SK Gupta [00:19:58] There are different dimensions to it, and it all depends upon what people are trying to do. Let's just analyze one at a time. In terms of cost, if you can speed up your operation as a result of using robots, then you can usually gain some benefits on costs. Also, if you do a holistic analysis, and there are components of the work which may cause injury to humans or pose some risk to human health, then there is a gain in terms of costs. Also, if you can run a multishift operation, again, usually, that can have significant impact on costs. There are different facets through which the point robot might give you a cost benefit, but it may or may not happen, depending upon the application. Now, if you think in terms of productivity, again, if you analyze it, the robot's ability to work much faster than the human... In some applications it's possible robot can actually be significantly faster than human. In that case, you can gain productivity benefits. In terms of quality, it usually comes as a result of consistency. Robots are significantly more consistent, and therefore, your variability in your production reduces. Whether or not these benefits are going to be present depends to a great extent on applications.
Gregg Profozich [00:21:19] I think those are some great points, SK, the whole concept, I think you were alluding to, of lights-out operation. I currently run a one-shift operation, but I have a robot now, and I have a long product run. I queue up the robot at quarter to five; I turn it on; it runs for eight hours; I go home and have dinner. We come back the next morning, and quality parts are sitting there. Overhead is being absorbed, production is happening. But marginal costs are pretty low at that point. Those are the things that we can get into when we use a robot and can get into creatively using it when it can produce quality consistently. Craig, anything to add to those — the cost, quality, and productivity benefits?
Craig Schlenoff [00:21:52] Just one thing. A lot of times people ask, "What is the return on investment? How much is it going to cost for the robot, and how long do I have to keep it running in order to make my money back?" As SK mentioned, it's highly dependent upon the application. There's robot manufacturers out there that claim that you can get your return on investment in two years. I can't verify that or not verify that, but that's what they claim. How much they cost is really highly dependent upon your application. There's the costs of the robot itself, 25K to 60K, usually in that ballpark. But then there could be other things that go along with it, depending upon your application. You mentioned before the end effectors or the grippers. If you can use the standard end effector or gripper, there's little to no cost there. If you need to come up with a specialized one, there's additional cost with that. At times you might need some infrastructure around it for safety or workflow. Now, if you have the cobots, as we talked about before, that's less of a concern. But if you can't use the cobots, then safety becomes an issue. Building fences using floor space take time and take money. Then there's questions of how easy is it to integrate. That depends, also, on your application. There are robot integrators out there that will be more than happy to integrate the robot into your factory floor, but that costs money. There's some that maybe what you're doing is simple and straightforward enough that you can take it out of the box and do it on your own. Some claim that you can do that in as quickly as four hours. Again, it depends on the application. As SK mentioned, there's lots of variables. Just some things to consider.
Gregg Profozich [00:23:16] Application dependent ROI. The cost of the robot. I can get a fairly low-cost collaborative robot with a standard end effector, and my cost is going to be pretty low. Integration is going to be mainly on the setup, because cobots are largely user-programmable. I've seen demonstrations where it's almost like hit the macro button. Press the red button; manipulate the robot with your hands; make it move through its motions; hit the red button again, and it's saved all the moves. Now you can just repeat that. Add in the safety indicators. If I'm doing a palletizing application, and I'm picking up one-pound boxes and putting them on the pallets, that's a whole lot different robot — payload, strength, safety requirements — than if I'm picking up 100-pound cases. That robot doing 100-pound cases is probably going to need safety cages and those things. There's considerations based on the application. It gets into all those variables and how they come into the total cost picture. If you were running a small or medium-sized manufacturing business, how would you go about developing a roadmap to identify and implement robotic solutions in your manufacturing process? Would you start in a particular place in the company? Would you start in a particular operation? Are there some standard off-the-shelf tried-and-true proven applications of robots? Where would you go?
Dr. SK Gupta [00:24:23] Start small. Start with applications which require the least amount of integration. Begin with things where you can use standard off-the-shelf robots, and standard off-the-shelf grippers, and standard off-the-shelf sensors with very simple integration. Start small. Make sure your workforce is ready. Sometimes the challenge is that you have deployed robots, and if you start having even slightest amount of challenges, you're not fully utilizing robots. Your robots are sitting idle, which then again is going to create significant challenges in terms of your ROI. Then build the right culture. Another challenge that often people face is that when people start introducing robots, if the culture is not right, and people are feeling threatened by the robots, adoption is not easy, because then people are always finding fault as to what robots are not doing. That just creates, again, challenges for you. You got to start small, make sure you have the right training for your workforce who's going to be interacting with robots or maintaining them, and have the right culture, when people see robots as helping your business, and creating new opportunities for you, and helping you grow.
Gregg Profozich [00:25:40] Great points, SK. Crawl, walk, run, I think is what I hear you saying. Start off with something very simple; get your people used to it. Because if your people don't accept the robot, a more complex one is just going to be harder to implement, and harder to get proven, and harder to be organizationally accepted.
Dr. SK Gupta [00:25:53] Yeah. We have heard those horror stories from many people. They picked an operation; they deployed an $800,000 cell; cell didn't work. Then when they call a system integrator, the cost of repair was $250 an hour. Therefore, they could never use it. Now they feel that robotics is useless.
Gregg Profozich [00:26:13] Yes. If we start small and start building the organizational acceptance. Right now I work on the production floor, and I'm a machine operator. A year from now if I see myself as a robotic technician and machine operator, I'm probably in good shape. If I still see myself as a machine operator who is in competition with the robot, we've got a problem. That organizational acceptance and building the right culture piece, I don't think it can be underestimated. There have been a few automation integrations that I've been part of where a vision system is supposed to check quality, and it does such a great job of checking quality that suddenly there's chewing gum on the lens, because the operators downstream are working harder than they ever did before. Those organizational acceptance issues type of thing, you have to have buy-in from everybody on the floor. We talk a lot about the technology so far, but that cultural element, that organizational acceptance piece, the change management piece, I think, is an incredibly important dimension. Craig, anything to add to what SK had said?
Craig Schlenoff [00:27:04] One thing that might be useful, especially for acceptance of people within the factory, is to focus the robot on the jobs that people don't want to do, assuming that aligns with what they don't want to do — things that are boring, that aren't safe, that are monotonous. If you do that, and maybe people realize they don't have to do that task anymore, that's also a nice selling point for having the robot there. Then the other comment is there's two ways of looking at bringing robots, I think, for businesses, one of which is I want to do what I'm currently doing, and I want to do it better, and faster, and cheaper. Then there's the second stage, which is I want to grow my business; I want to grow into other areas; I want to be able to do additional things. If you have a plan for how you want to grow, that can help you to better identify if a robot will work for you and what tasks it will work for. They're similar, but they're different in the sense that you're not just doing what you're doing better, but you're trying to figure out where the company wants to go and seeing if the robot aligns well with it.
Gregg Profozich [00:27:56] Great ideas. From an operations perspective, how long does it take to program a robot, and what skills do I need to have? It's probably a different question depending on the type of robot I implement. Let's talk about the different dimensions there. How long does it take to program, and what skills do I need to have?
Craig Schlenoff [00:28:11] The answer to this question, like pretty much all the other questions, is it depends. It depends on how sophisticated of an operation you want the robot to do. As I mentioned earlier, some robot manufacturers claim that you can get the robot from the box to integrate it into your shop floor within four hours. That's true, I'm sure, but for more simplistic types of operations, and you don't need a robot integrator or anything to do that. But there's things that are much more complicated. If you want to do things that require special end effectors that need high accuracy, some of the things that SK was talking about before, then it's going to take more time to get it up and to get it integrated. There's really three steps with a robot. It's programming it and getting it up and getting it working. That takes time. There is operating. Once you have it programmed, running it and making sure it's doing the right thing and it's continuously doing the right thing. There's maintaining it. Just like every other piece of equipment that's on the factory floor, every other piece of machine, things wear down, things sometimes get out of skew and don't work exactly the way that they're supposed to. There's time that goes along with maintaining it. All three of those things are necessary. As I mentioned earlier, robot integrators are available to help. This is their specialty, and then they do that stuff, but it costs money. It really depends upon the sophistication of your application and what you're trying to do. The robot manufacturers have gone out of the way to make the robots as user-friendly as possible. But what skills do you need? Certainly, familiarity and comfort with equipment, which I imagine would happen to most places. They've tried to get away, for the most part, from programming skills, even though coding is useful and interesting. There's ways, as you mentioned earlier, to be able to teach a robot to do something without typing a line of code. But sometimes it gets more complicated, and you need those skills. It really depends on how you want to apply it.
Dr. SK Gupta [00:29:49] Another dimension which is increasingly happening is sensor integration. People are interested in getting the sensor data to the robot, and also, sometimes you have multiple different robots or you have a robot being coordinated with a machine. Then you get into PLC, program and logic controllers, as well. Depending upon how complex your robotic cell is, whether it is going to be including external sensors, whether it will include coordination with other robots or machines, this can get complicated. But the simplest robot where you just demonstrate by holding the robot the motion it needs to perform, and record, and replace, quite simple. It can be anywhere from a few hours to a month, depending upon how complex things can get.
Gregg Profozich [00:30:32] Again, back to our crawl, walk, run approach. Pick the simplest thing, get some buy-in, learn those skills along the way. One of the concepts, I think, in the book Lean Robotics, is do exactly that. Take a very simple application, but take your first robotics implementation as the first of a series of them, where you're going to learn internally, and start building that internal capability to understand where to apply it. There's always going to be systems integrators out there to help. But generally speaking, you probably want to be in a position where you can be deciding what you want to do based on your understanding, your expertise, not relying completely on someone else's. You start that journey with a simple application of a robot, and then understand the technology. As you understand the technology and application better, you can expand it into various other areas of opportunity within your operation. Let's talk a little about reliability. How reliable are robots? What happens if they break down? How do I fix them? Do I need somebody on my staff? Talk about those as concepts.
Craig Schlenoff [00:31:26] A robot is a machine similar to the drill and the other machines that you have on the shop floor. They work, for the most part reliably, but they break sometimes. When they break, they can break in multiple ways. Sometimes there's a person in the shop floor that says, "Oh, I've used this before, and the drill bit broke," or whatever it happens to be, "and I know how to fix this." They fix it, and it's done. Sometimes it's much more complicated, and you need to call the manufacturer and have the manufacturer come out. If it's under warranty, that's great. If it's not under warranty, there's usually a cost associated with it. But it's the same thing as other types of machinery that's on your shop floor. Now, obviously, robots are a little bit more sophisticated, and there's more ways that they can break. Maybe there's a slightly higher probability that they're going to break before the machinery that you have. But it's the same type of thing. Fix it where it's obvious what needs to be fixed, and if not, call in the company and have them fix it.
Gregg Profozich [00:32:15] The skills can be basic mechanical skills, I hear you're saying. Replace a drill bit type of thing. If it's a mechanical skill and something obvious, you can fix it. If we don't know why it's adding a step into the operation that wasn't in the program before, it's probably going to require the expert who programmed it in the first place type of thing. Right?
Craig Schlenoff [00:32:29] Right. There is the mechanical side, which is exactly what you said. But robots are ideally somewhat more intelligent — hopefully, more intelligent — than some of the other machinery. There's a lot more to the software side, as well. If there's problems with the software, then you need to either have somebody on staff that's a whiz programmer, which I know many of these small manufacturers probably don't have, or you got to call in the company.
Dr. SK Gupta [00:32:50] Building upon what Craig said, arms, in general, tend to be fairly reliable. When you buy an arm, you can buy an extended warranty or service program, where they will support any failure in the arm within a certain period of time. But then the failure could also happen at your processing unit. If you're doing something that your gripper fails, or something else happens, your chances that failure would happen where you're making contact with the world are much higher than your failure chances with the arm itself. You have to take a look at where the failures are happening. Is it at the center level? Is it at the arm level? Is it at the end effector level? Is it your end of arm tooling that you're using, this feeling, or something has gone wrong with your program that you were using to run it? Depending upon the failure, you need different interventions.
Gregg Profozich [00:33:41] I think I hear you saying, between the two of you, that some things, a simple mechanical thing, the screw's obviously laying on the floor right below the place where there was a screw hole empty, I could put the screw back in and fix that. The simple mechanical stuff we can do internally. Some stuff is going to require outside help. But there are experts either at the machine manufacturer, the end effector manufacturer, or the integrator who could help bridge that gap and support you in that way. Buying a robot you're not alone, that there's an ecosystem that can support long-term maintenance and repair. We talked a little bit about this, but how do I ensure that it's safe for my employees to work around the robot?
Craig Schlenoff [00:34:16] I think this is getting back to the discussion, as mentioned, that we had earlier. If you buy a cobot, a collaborative robot, they are designed to be safe to work around people, because they conform to the standard specifications that were mentioned earlier. They're certified to do so. It's not easy for a manufacturer to call their robot a cobot. They have to make sure they meet these specifications and show that they meet these specifications. There's the implicit understanding that they will be safe around people because of the things we talked about earlier: because of limiting the amount of force, by going slower when they get close to the people, by the fact that most of them are carrying less weight, there's less force momentum to be able to hurt people if they knock into them. If you're able to use cobots for your application, you can feel a reasonable level of comfort that they're going to be safe around people within the shop floor.
Gregg Profozich [00:35:02] Okay. That's for the cobot case. How about the robot case?
Dr. SK Gupta [00:35:04] Robot case. If you're not keeping them in the cages and if you are not ensuring that absolute human separation, you can always have a system integrator perform a risk assessment and give you a report as to what the modalities are, what risk it poses. Then you can develop mitigation strategies to ensure that those risks remain manageable.
Gregg Profozich [00:35:31] If we do a safety risk assessment on a regular robot and it has a force and a velocity that would be dangerous to humans to be around, they can put it in a safety cage; you can put up a light curtain. There's many different ways to interrupt its processing, to make sure that it's a safe environment. Let's talk about some specific cases. For example, hypothetically, my company makes very customized products. Are robots good for that type of application, when I've got many different things that are very custom?
Dr. SK Gupta [00:35:56] It depends upon where the variability is coming from. If you are, let's say, milling different kind of parts and if you're using robot for machine tending, then absolutely you could go ahead and deploy it. That's not an issue. Now, if you are going to be, let's say, performing some processing operation, and you're saying that okay, I'm going to be welding different kinds of structures, and I want to have a welding robot, things will be much more harder than that case. It all depends upon where the customization in the product is coming from and what tasks are you trying to perform. Welding different geometry is very different the way a robot can move. On the other hand, if you're trying to do machine turning or you're trying to inspect a part, then even though there's a variability in the geometry, but the robot motion may be very similar in nature, and therefore, it will be much, much easier to deal with that variability in the context of that task versus a processing task where a robot has to perform a fundamentally different task. Certain kinds of variability is very easy to deal with; a certain kind of variability is challenging. You have to take a look at what kind of tasks we're talking about.
Gregg Profozich [00:37:13] It's more task dependent than product dependent or even business model dependent. Anywhere there's repetition of the same path, the same payload, et cetera, et cetera, those are applications of robots regardless of what I'm making.
Craig Schlenoff [00:37:25] Just to build off of what SK said, likening it again to a large manufacturer like a car manufacturer. If you look at the cars that are coming out of the plant, you can imagine all the options you could have on the car. Any two cars are likely not going to be exactly the same, but a lot of the fundamental processes to build those cars are the same. You still need to put a roof on it, you still need to put tires on it, even though the entertainment system might be different or whatever else. Just building off of SK's comment, depending where that customization is, robots still could be very, very useful, which is maybe not for the details of where customization occurs.
Gregg Profozich [00:37:57] Okay, makes sense. Let's talk a little bit about the future. Where do you both see robotics in the manufacturing sector over the next decade or so?
Dr. SK Gupta [00:38:06] Sensing technology is getting cheaper and cheaper. Therefore, you will see in a robot deployed a lot more sensors. Once the robot have the sensing modalities, then, obviously, they can do much more complex tasks than what they were able to do before. We are also seeing a lot of advancements in AI and machine learning. Those are then giving robots the ability to basically reason. Once the robot can reason, then, again, you can push them in new application area where you're not able to see them today. You're also seeing a lot of improvement in user interfaces. Companies are now offering a smart pendant. People are talking about augmented reality. There are lots of new modalities through which humans can interact with robots. All that, again, makes it much, much easier for people to program not just a robot but also probably a cell. We're also seeing new technology for human safety, because now you have much more reliable technology for tracking human. You can track human much more accurately. Therefore, in future, hopefully, some of your traditional style industrial robots which are not considered human safe, you can actually interact with them in a very, very different mode than what you're allowed to do today. Sensing, AI, next-generation user interfaces, and human safety are all different technology which are coming down the road. All of these will have huge impact in terms of applications where robot can be used and also the overall cost of acquiring a robotic cell or a robotic solution.
Gregg Profozich [00:39:44] Craig, your thoughts?
Craig Schlenoff [00:39:45] Another way of looking at this is that the robot needs to do at a minimum three things. They need to be able to sense the environment; it needs to be able to think, and it needs to be able to do. SK talked about the sensing and some of the other areas, also, but that's obviously the first step. That, for the most part, if the robot can't see... Likening it to humans, if a human can't see the environment, it's very tough to do things that are productive in it unless it's very well-scripted. I think even sensing technologies are starting to get better and giving robots more knowledge in order to do that kind of stuff. The thinking is what is my biggest area of interest. Robots are starting to get more intelligent. We're not going to see C3PO or R2D2 anytime soon, but they're able to do things that they weren't able to do before. One of the things is... We're not there, but we're getting closer to robots figuring out how to perform a task without the person telling them what to do. For example, by feeding in I want this to be my end state, I want you to put this ball in this hole over here. Here's where the ball is, and here's where the hole is. Figure out a process in order to do that. I don't have to grab the robot's arm, and move it around, and tell it where to go. It can figure out the plans that it needs to go through in order to get from the current state of the world to the state of the world that you want it to be in. We're taking steps in order to get closer to doing that. By doing that, we're going to do things such as greatly decelerate how long it takes to get a robot up and running, because you just need to give it the state of the world, and it can figure out how to perform the tasks that you want it to do. Then the do side, obviously, builds from that. These are planning types of applications and actuation. Robots are getting more capable in what they can grasp and what they can move around. I think we're seeing a lot of work done there, as well. Everything that SK said I agree with. Just providing a little bit more context. I think the thinking side is the side that we're really going to see a lot of the benefit from moving forward, very comparable to a human. Then one quick story is another factory that I went to... I seem to go to a lot of factories. One other factory that I went to, we were walking around. It was a large factory. As the guy was showing us around, we heard sirens go off and lights flash. We said, "What's going on?" He says, "Excuse me for one second." He walked over to one of the robots where the lights were flashing and the sirens were going off. He pushed the e-stop button, and he grabbed the part that happened to be in a light curtain. Essentially, there's a light curtain to make sure that that part was in the right general area for the robot to pick up. It didn't quite get into that area. The laser was broken to show that it wasn't in that part. The person went and took that part, slid it over about three inches so that it was in the area in which it was supposed to be, closed the door, pushed resume, and the robot was able to continue. Something as simple as a part being a little bit off-skew caused the factory at that point to shut down for a moment. The person had to go in and fix that. Those are things that... Again, doing the analogy with a two-year-old. The two-year-old could probably figure out I need to move something over in order to get that to work. Part of that thinking, of understanding where things are going wrong and what the robot itself can do in order to fix it without setting off the alarms, I think, is where we're going in the robotics world.
Gregg Profozich [00:42:49] We've covered an awful lot of ground today. I thank you both for all of your insights, and your ideas here, and sharing with us. I'm going to try to do a quick summary here. In terms of if I was a small manufacturer, the things that I would be thinking about in considering robots, are they right for me, is first start off with the dirty, dull, and dangerous jobs. Where can I use a robot that is going to replace operations and tasks that people don't like to do, places where I have high turnover, places where I have the risk of or a frequency of lost time, injuries, and workers' compensation claims, those kind of things? Start there to really recognize the benefit and really get the full value of a robot, because the robot's not going to suffer any of those injuries. Robots are great for flexibility and consistency in operation. They allow for innovation. They can do things with high quality and high precision. Repetitive tasks with low variability is the sweet spot as a good starting place. Then you can go somewhere else with your future robotic implementations from there. Robots are making advances. I think we just wrapped up in the last part of the conversation some of the advances in the future, but they're likely never to replace humans. Even with all the advances in AI, vision, sensors, et cetera, getting to the point of being equal to the skills of the human two- or three-year-old is quite a challenge. Do things that are repetitive that are low variability, things like material handling, inspection tasks, machine tending, palletizing, operations where there is a similar motion path. Those parts or those things that are being moved by the robot or manipulated by the robot should have a similar payload, if you will, similar weight and size characteristics. Look for larger lot sizes in my operation. If I'm a small manufacturer looking at this, the longer the production run, the more chance that the robot will be able to deliver quality consistently and help me with the return on investment. Also, opportunities for that lights-out operation that I think SK mentioned. If I can set up the robot 10 minutes before the end of the shift and turn it on, and let it run for eight hours, and go home, and have quality parts produced, that is all product that falls straight to margin. It's a great opportunity for getting to ROI. Cobots versus robots. Cobots have some safety features built-in in terms of force limiting, and speed and separation monitoring that they have to do. If you buy something that is certified as a cobot, you have those pieces built-in. If you buy something that is not, it's a robot, and there'll be some safety considerations, but there are ways to mitigate risks. Integrators can help with those risk mitigation plans to help get safety cages, or light fences, or whatever the appropriate safety equipment in place to make sure that it's a safe environment. Productivity. Robots can work much faster than humans in some cases and can deliver more consistent product, which impacts the quality. The ROI, though, is going to be dependent on the cost of the robot, the end effector that's used — is it standard or is it custom — the integration, and the safety consideration. It's not just the cost of the robot to consider, but you have to look at all four of those things. But there are some simple, almost packaged solutions that robot makers and integrators have put together that can solve some of those problems and get to a very reasonable cost. If I was starting out, the advice I heard was start small. Use the standard robotic arm; use a standard gripper; use a standard software; and make sure the workforce is ready. Make sure they understand why we're doing this and how it's a benefit to them. Third piece of that is really build that culture. Make sure you gain the organizational acceptance. Do the change management work to make sure that it all comes through. In terms of programming and maintaining the robots, the programming is going to depend on the application. Some cobots and robots have very user-friendly interfaces that don't require writing code; some are more complex. It's going to be application-dependent to see what that level is. Programming, operating, and maintaining the machine. Like any other piece of equipment, those are the steps. It has to be set up; it has to be operated; it has to be maintained. In terms of skills for that, programming, if it's user-friendly enough, your internal people can do it or can learn it fairly easily. You may need an integrator or a programmer up-front to get started. In terms of maintaining, some tasks are going to be easy and mechanical to fix for your maintenance crews; some are going to require more advanced skills. It's going to be a trade-off and really be dependent upon the individual application situation. In the future of robots, sensor technology is becoming less expensive. That in combination with artificial intelligence and machine learning is increasing a robot's ability to learn, and to think, and to almost reason. As those advance the application of robotics and the ability to apply to a wider set of problems to solve is going to grow. Did I miss anything in that summary? Is there anything else you want to add or correct? Did I misstate?
Dr. SK Gupta [00:47:17] Pretty good.
Craig Schlenoff [00:47:18] That's an amazing summary. Well done.
Gregg Profozich [00:47:21] Thank you both for being here. I really, really appreciate you being here today. I learned a lot from this conversation. I want to thank you for joining me and for sharing your perspectives, and insights, and expertise with me and with our listeners.
Craig Schlenoff [00:47:33] My pleasure to be here.
Dr. SK Gupta [00:47:34] Truly a pleasure to participate.
Gregg Profozich [00:47:36] To our listeners, thank you for joining me with this conversation with Craig Schlenoff and SK Gupta in discussing Robotics & The Small Manufacturer Part 1: The Evolution of Robotics Technologies and Manufacturing. Thank you so much. Have a great day. Stay safe and healthy. Thank you for listening to Shifting Gears — a podcast from CMTC. If you enjoyed this episode, please share it with others and post it on your social media platforms. You can subscribe to our podcast on Apple Podcast, Spotify, or your preferred podcast directory. For more information on our topic, please visit www.cmtc.com/shiftinggears.
CMTC is a private nonprofit organization that provides technical assistance, workforce development, and consulting services to small- and medium-sized manufacturers throughout the state of California. CMTC's mission is to serve as a trusted advisor, providing solutions that increase the productivity and competitiveness of California's manufacturers. CMTC operates under a cooperative agreement for the state of California with the Hollings Manufacturing Extension Partnership Program (MEP) at the National Institutes of Standards and Technology within the Department of Commerce. For more information about CMTC please visit www.cmtc.com. For more information about the MEP National Network, or to find your local MEP center visit www.nist.gov/mep.