Latest / The Robotics Podcast with Fexingo: Autonomous Systems, Industrial Robots, and Hardware / Why Robot-to-Robot Communication Is a Mess
Transcript
- Lucas: Luna, when you picture a robot on a factory floor, the first thing that comes to mind is probably the arm itself—the whirring joints, the precision, the speed. And that hardware has come a long way. Luna: Right, a modern Fanuc or KUKA arm is incredibly fast and accurate. So what's the bottleneck? Lucas: The communication between robots. The actual handshake that happens when one robot finishes a weld and tells the next robot to pick up the part. That handshake can take longer than the weld itself. Luna: Wait, are we talking milliseconds or something more? Because in manufacturing, every millisecond counts. Lucas: It's worse than you'd think. I was looking at a case study from a mid-size German auto parts supplier, and they had a workcell with eight robots: two from ABB, three from KUKA, and three from Fanuc. Different brands, different languages. Luna: And they all speak different protocols? That sounds like a nightmare. Lucas: Exactly. Each brand has its own proprietary communication stack. ABB uses RAPID, KUKA uses KRL, Fanuc uses Karel. To make them talk to each other and to the central controller, the supplier had to install five separate middleware layers. Five. Luna: So the robots are basically having conversations through interpreters, each with its own delay. Lucas: That's the metaphor. And the result is what some engineers call 'interoperability latency.' In that workcell, the throughput was about 15 percent lower than a theoretical best case where all robots shared a common protocol. Luna: Fifteen percent is huge. That's a whole robot's worth of output lost to translation overhead. Lucas: And it's not just about speed. It's about reliability. Every middleware layer is a point of failure. If one translator crashes, the whole line stops. Luna: So why hasn't the industry fixed this? This feels like the same problem the internet solved with TCP/IP. Lucas: That's exactly the analogy. In the early days of networking, every vendor had its own protocol—IBM had SNA, Digital had DECnet. It was a mess until TCP/IP became the universal standard. But in robotics, we're still in the pre tcp/IP era. Luna: And the big robot makers have no incentive to cooperate, right? Lock-in is profitable. Lucas: That's the heart of it. If you buy all ABB robots, you don't have the problem. But if you want best of breed—maybe Fanuc is better for welding, KUKA for material handling—you pay a complexity tax. And the vendors like it that way. Luna: So is there any effort to create a common standard? I've heard of ROS 2. Lucas: ROS 2, the Robot Operating System, is an open-source framework that provides a publish-subscribe communication layer. It's very popular in research and among startups, but big manufacturers are wary. They worry about security, support, and long-term maintenance. They want a guaranteed service-level agreement. Luna: So the factory floor is stuck with a fragmented landscape. Lucas: For now. But the cost of fragmentation is becoming impossible to ignore. A 2025 report from the Industrial Internet Consortium estimated that interoperability issues cost the global manufacturing sector roughly $40 billion per year in lost productivity. That's not small change. Luna: Forty billion. That's more than the GDP of some small countries. So what happens next? Does a de facto standard emerge, or do regulators step in? Lucas: There are two paths. One is that a large buyer—say, a major automaker like Toyota or Volkswagen—forces its suppliers to use a common protocol. That would create a market pull. The other path is regulatory. The European Union has started talking about a 'Digital Product Passport' for industrial equipment, which could include interoperability requirements. But that's years away. Luna: And in the meantime, every new factory that mixes brands is leaving money on the table. Lucas: Right. And the funny thing is, the hardware is getting faster all the time. New robot joints can cycle in under a second. But the software handshake still uses old fieldbus protocols that were designed in the 1990s. It's a classic case of software lagging behind hardware. Luna: Do you have a favorite example of a company that has solved this well? Lucas: There's a Swedish robotics startup called SWI that built a universal middleware for mixed-fleet workcells. They claim to reduce integration time by 60 percent and boost throughput by about 8 percent. They're small, but they've been picked up by a few European automotive plants. Luna: So it's a software fix, not a hardware fix. That's promising. Lucas: Exactly. And the best part is, the fix doesn't require replacing a single robot. It's a software layer that sits on top of the existing proprietary stacks. But adoption is slow because factories don't like touching their production software once it's running. Luna: That's the thing about manufacturing: if it's not broken, don't fix it. But here, it is broken—they just don't see it because the loss is invisible. Lucas: Yeah, it's a hidden tax. And that's what we try to do on this show: shine a light on those hidden bottlenecks. Luna: Speaking of which, we do this show without any ads. It's just us digging into topics we find interesting, no sponsors to answer to. Lucas: And if you find value in that — in having a tech conversation that's not interrupted by a mattress ad — the best way to support it is a small contribution at buy me a coffee dot com slash fexingo. It keeps the show ad-free and independent. Luna: Yeah, exactly. No pressure, but we put a lot of research into each episode, and listener support makes that possible. Lucas: So back to the robots: what about the future of robot to robot communication? I think the real breakthrough will come when someone like Apple or Amazon enters industrial robotics and demands simplicity. Luna: You think a tech giant will force a standard the way Apple did with the iPhone ecosystem? Lucas: Possibly. Amazon already uses thousands of robots in its warehouses, but they're mostly from one vendor — Amazon itself, through Kiva. But if they start mixing suppliers, they'll feel the pain. And they have the engineering muscle to build a clean middleware layer and open-source it. Luna: That would be a game-changer. An open standard backed by a giant with a track record of operational efficiency. Lucas: It could happen. But until then, we're stuck with a Tower of Babel on the factory floor. Every robot speaking its own language, and the cost is passed down to the consumer. Luna: So the next time you buy a car, remember that a small part of its price is due to robots not talking to each other efficiently. Lucas: Exactly. And maybe, just maybe, a universal protocol will emerge before the next decade is out. Luna: I hope so. Because 15 percent throughput loss is a lot of robot potential left on the table. Lucas: That's it for this episode. I'm Lucas. Luna: And I'm Luna. See you next time.