The major shift toward humanoids podcast transcript

Dale Tutt: Welcome to the Industry Forward Podcast, the show where we examine industry and technology trends with the help of various experts from around Siemens and beyond. My name is Dale Tutt. I’m the Group Vice President of Industries at Siemens Digital Industry Software and your host for today’s podcast. 

And today with me, I have Rahul Garg, who joins me from our Industrial Machinery team. Welcome, Rahul. 

Rahul Garg:  Hi, Dale, very excited to be here. So just by way of quick introduction about myself, Rahul Garg, Vice President for Industrial Machinery Vertical, focusing on software strategy and Siemens Digital Industry software. 

I’ve been with the company for around 25 years, but what I sometimes call as a startup within the large enterprise of Siemens, and prior to that, have worked for three other startups. My first startup actually was around robotics as well, where we were building optical jukeboxes for storage of large-format drawings and other things. Yeah, excited to be here today with you. 

Dale Tutt: Welcome to the show, Rahul. And today you went in this direction already, talking a little bit about your experience as a startup company working with robots. But today we’re going to discuss humanoid robots and the various challenges that come with designing, building, and implementing them in a factory. 

You and I were at the Consumer Electronics Show here last January, and it was amazing to me that everywhere we looked, there was humanoid robots. And they were all doing these demos and some of them were playing piano and a lot of them were doing karate and boxing demonstrations. 

But I guess, robots were everywhere anyway, autonomous lawnmowers and all kinds of robotic devices.  But before we dive into what I hope is going to be a very exciting and informative discussion, maybe we should talk a little bit about what a humanoid robot is.  A lot of people have a lot of different visions of what these are from the movies and from other things. 

There’s a lot of different interpretations of what a humanoid robot is. We talk about humanoid. Are they always bipedal? Do they always have hands and arms and a head? Versus, say, an AGV or an autonomous ground vehicle that is, you know, being able to move out the shop floor freely. Sometimes these AGVs might have a robot arm on them. And for distinction for our listeners, what is a good definition of a humanoid robot versus an AGV? 

Rahul Garg:  First of all, what I would say is a humanoid robot is a mobile robot system, right? It is mobile. It’s designed to have a human-like body structure, I would say, could be legs, certainly, but it could be a wheelbase as well, have arms, a torso, often a head. And the main intent is that it can operate in an environment designed for humans without requiring infrastructure changes. That’s a big piece, I would say. 

Typically, they have multiple axes for manipulation. They are battery powered, a lot of dexterous hands for getting into different types of motion. Integrated perception, vision, depth sensing, tactile sensing, and they are software controlled, and then they’re battery powered as well, right? So that they have that ability to move around a lot more freely. Those are some of the key things I would say from a humanoid perspective, right? 

From an AGV or an AMR, right, which is like your automated guided vehicle or autonomous mobile robot, right? They are primarily designed for material transport. They’re not designed for manipulation. They’re not designed to do work of a human. Their main objective is take stuff from one spot, then move it to another spot. They are typically wheel-based. They don’t have legs for sure. 

They may have arms, they may not have arms. The navigation is marked with LIDARs, markers, slam. And then they are primarily optimized for logistics, like typical use cases are like pallets, cart movements, warehouses. They are obviously lower cost than humanoids, simpler mechanics, the degrees of freedom that they would have is probably around six, seven at the most. The degrees of freedom that a humanoid may have could be anywhere from 20 to 50. So that becomes a big factor in trying to differentiate or at least help understand the differentiation between the two. 

Dale Tutt: I think there’s a good distinction there because a lot of times the AGVs, as we traditionally think about them, are for material transport. Occasionally you’ll see an AGV with what looks like a traditional robot arm. You know, when I think about a humanoid versus the traditional robot is that the traditional robot is often is fixed to the floor and the arm has, you know, maybe limited ranges of motions and very specific tasks. And the humanoid, the example that you use, the humanoid robot, it could be like on an AGV and it could be a torso with arms on an AGV. But it has that freedom of motion, the motion that like you would expect for a humanoid robot. 

You’ll go to these restaurants and you’ll see things that picking up the trays or delivering food. And to me, that’s more of an AGV because it’s not actually taking the food off of the robot and putting it on the table. It’s bringing, it’s transporting the food to your table and you can take it off or you can put your dishes, your dirty dishes on it. And so they try to make it like have a smiling face or something on it. That doesn’t feel humanoid to me. That feels like an AGV because it’s just moving stuff around, but it doesn’t have that range of motion and the decision making, I would say, that you might have built into a humanoid robot. Is that a fair statement? 

Rahul Garg:  That’s actually a very good example. The day I would say servers have the ability to actually pick up the plate from your table and put it into their plate holding container or in their tummy or whatever. That’s when they would transition from being a sophisticated AMR to becoming a humanoid. And to your point earlier, right, of how does this all compare to a more traditional industrial robot, right? 

Those are typically fixed base. They perform the same task repeatedly. And they’re typically electrically connected. They don’t need to have a battery pack. They’re bolted to the floor and they’re performing that high repeatable motion. They’re pre-programmed for that task. And they are very high precision, right? 

I mean, that’s been one of the biggest values. They have tremendous capacities to lift heavy payloads. They can pick an entire vehicle up, weighing a few tons and move it around as necessary. So obviously a lot of strength and all those things. Sometimes I think it’s more because I mean, it’s even the products have to be designed around, in a traditional robot environment, you know, the products have to be designed to work with those exact robots. 

And so now we’re seeing this form factor where they’re taking on a humanoid shape and either shape of the arm. And you might be able to question why you would have legs and something that looks like a head, really does take on the shape of a humanoid. Why do you think that there’s so much emphasis on having them look humanoid and have that level of functionality versus maybe a little simpler or less humanoid looking? Where do you see that the advantage of the shape there? 

Rahul Garg:  One of the biggest value propositions and the reason why there’s such a big focus on this form factor is because it enables a very easy drop-in and a replacement for existing human labor, right? And especially in brownfield environments. If you think about it, our world that we live in over the last three, four centuries, it’s been designed and engineered for a man-made world, right? 

For us as humans to function in it, everything around us, our homes, our factories, the way we do things, it’s all around our form factor. So, that’s what the humanoid is looking to do, is basically have the ability to fit in that, what I would call as the brownfield environment, and being able to fit straight into that infrastructure without significant changes, and the ability to interoperate with the human right. That’s kind of the biggest holy grail, right, of this whole process. 

Dale Tutt: How do you think companies are going to balance humanoids with AGVs and maybe even traditional robots? I mean, because it does feel like there’s a place for all of the above, depending on what the operations are. But what are your thoughts on that and what have you heard from some of our customers? 

Rahul Garg:  That’s a great question, Dale. It’s actually the billion-dollar question because as these new greenfield factories are being designed and built. Every company that is looking to make a big investment in one of these is looking at how a humanoid can change that factory in a few years, right? Because the typical life of a plant is definitely more than 10 years, right? 

So these things are at the cusp of moving from demonstrations to production usage. Last year, there were some 25,000 of these that were shipped, these humanoids. This year, the number is close to globally across all the companies. It’s close to 100,000, 150,000. It’s scaling at a very rapid pace, very, very rapid pace. The only thing that’s holding it back is the production capacities of these companies. The plant of the future is not going to be the plant of today because the flexibility and dexterity is going to have a big impact. 

What companies are doing is obviously they cannot wait for five, 10 years to wait for these things to become the big holy grail answer. In the meanwhile, they are beginning to design factories to be a lot more modular. By modular, I mean have the ability to bring in a human to perform a certain task and then having the ability to replace that human with a humanoid. In fact, certain companies that are building these humanoids, are already thinking about that, right? Having a human build that humanoid, but in some time having a humanoid build certain parts of it and over some stage of time, bring in a lot more of that automation. 

Dale Tutt: Wow. Fascinating by the numbers that you threw out, like how many we delivered last year, how many robots, humanoid robots are going to get delivered this next year. And you made the statement, they’re just limited by how fast the factories can design and produce them. 

So, you know, as think about the design challenges of a humanoid, I think about it a little bit sometimes at a macro level that obviously it’s got to be strong enough to do the work and it’s got to have the dexterity and the motion that’s required. But then I think about just even some of the technologies that you have to have battery packs that aren’t going to just burn out every two hours and you’re always at a charger and you got to have electronics and semiconductors that are designed, purpose-built for these robots and the software. And these become a classic software-defined product. 

As we’ve been talking today, one of the things that has really struck me that you’ve talked a little bit about, you bring the technology and you have the robots and you bring them into your factory. Then you also have to think about some of the safety considerations and everything else. And a lot of times when we talk about digital transformation, there’s a discussion around people, tools, and processes. 

And it’s not just about bringing new tools in like robots. It’s not just about changing the processes, but implementing it in a way that is actually is adopted by the people. Because you can kind of imagine, I can imagine that there’s a lot of resistance to seeing these kind of robots starting to show up in the factory floor working next to people. It’s going to continue to be a topic that you have to look at this holistically. It’s not just about jamming a bunch of robots in the factory and seeing how it goes. 

It’s really about thinking about how does this incorporate into an efficiently run factory situation. I don’t know if you’ve heard any conversations about that, those kind of topics, but it does seem that there’s an organizational change management piece that has to happen alongside the adoption of this technology into the factories. 

Rahul Garg:  Yes, it’s certainly a very important piece, Dale. In fact, we are building a whole digital thread around this topic, right? Which is getting the humanoid to coexist with the rest of the production systems that you would have in place, which starts all the way from the initial calibration training for that specific task that you’re looking at, trying to digitally figure out the human-humanoid machine interaction process, right? Trying to make sure it’ll work the way you are expecting it to perform. 

And then even getting down to the job execution, right? End of the day, that machine, that humanoid is performing a task that is controlled, or at least that’s defined in our manufacturing execution systems. So the manufacturing execution systems that have been written up and that have been used for a long time were written up and built with the human in mind. A whole bunch of work instructions that have been written up based on that. 

Now, how do you transform that for a humanoid to use so that as the job sequencing happens, they get the right instructions, they know what part to pick up, what screw to use, what torque to apply, making sure that they perform that task, and then they complete that task to the satisfaction, report the quality, and then go to the next task, go to the next job. That whole execution loop that was designed. Now it’s gonna be performed by humanoid, right? So, trying to map that piece out, making sure that the workforce that was there before, they may be needing to do something different now. 

How do you evolve the workforce in that context as well, and then how do you even provide feedback to your engineering, manufacturing engineering people and to your OEM that provide you to the humanoids so that they can make more changes or updates to the current that’s sitting on your shop floor. There are many different issues that need to be considered, and that’s where we are trying to put this all together in this context of this coexistence loop, which I think is going to be very critical for companies to understand. 

Again, helping them figure out how do we go from a simple proof of concept to now getting it into a production environment. 

Dale Tutt: 

Well, Rahul, it’s been a great conversation today, and I really appreciate your time and the discussion. And as we wrap up, we had a great chance to talk about what humanoid robots are and just as importantly, what they are not, and how humanoids are different than AGVs and traditional robots, and how they provide a different solution for automation in an existing factory situation. 

But also highlighted why humanoid robots can be so compelling, not because they look like they’re human, but because they are designed and built to function in an environment that was designed and built around humans. And this is what makes them such a great solution for companies with brownfield factories or with existing products and production processes. They are able to incorporate automation through the use of humanoid robots without redesigning their products or their production processes. And so this can be a game changer for these companies to add automation. 

And as companies are looking at greenfield situations, they’re also now starting to look at how do you incorporate different forms of automation, such as humanoids and AGVs and traditional robots alongside humans, to best optimize their factory situation. So it really has changed how companies are looking at their factories. 

But most importantly, we talked about the fact that this isn’t just a technology problem, that this is about people and processes and safety and change management and being able to bring humans and humanoids together on the shop floor in a safe manner, and how your digital solutions and your execution systems need to evolve in order to be able to support humanoid robots on the factory floor. 

So humanoid robots, they’re not a silver bullet, but they’re clearly moving from demonstration to real production use. And the companies and the factories that take advantage of it now and start thinking about it now and start laying that digital foundation with the digital twin and digital thread are going to be the companies that are best positioned to take advantage of this new form of automation in the future. 

So Rahul, once again, thank you for your time today and for sharing your insights. And thanks to everyone that listened. And if you like what you’re hearing and you’d like to learn more and hear more like this episode, please follow the Industry Forward podcast on your favorite podcast channel. So we’ll see you next time and thank you very much. Thank you. Goodbye.