Latest / The Robotics Podcast with Fexingo: Autonomous Systems, Industrial Robots, and Hardware / Why Robot Arms Still Cant Pick Up a Coin
Transcript
- Lucas: So you walk up to a vending machine, drop in a quarter, and the machine — after some clunking — spits out your soda. Simple enough. But imagine if a robot arm had to pick that same quarter off a flat counter. Turns out, that's still a genuinely hard problem. Luna: Harder than folding a fitted sheet? Lucas: In a different way, yeah. With a sheet, the challenge is flexibility and deformation. With a coin, it's the opposite — it's thin, rigid, smooth, and lies completely flush against a surface. There's basically no gap for a gripper to get under. Luna: So how do current robot arms handle it? Or do they just skip coins entirely? Lucas: Most industrial pick and place systems avoid the problem. They handle objects with some thickness or texture — boxes, bottles, circuit boards — where you can use suction cups or parallel jaw grippers. But a coin? A standard suction cup can't create a seal because air leaks around the edge. And a two-finger gripper can't pinch something that's flush with the table. Luna: Right, because there's nothing to pinch. The coin is essentially a thin disk sitting flat. Your fingers would just hit the table. Lucas: Exactly. That's the core geometric problem. And it's not just coins — it's any thin, flat object: a credit card, a piece of paper, a plastic lid. Those are everywhere in manufacturing and logistics, and robots still struggle with them. Luna: We've actually done a few episodes on paper and wet paper towels. But coins feel like a more extreme version — hard, slippery, metallic. Lucas: Right. So there's been some interesting research recently. A team at Stanford published a paper last year on what they call 'electrostatic adhesion grippers'. The idea is to use a voltage to create an electrostatic charge on a flexible pad, which then attracts the object. Luna: Like a balloon sticking to a wall? Lucas: Very similar principle. But tuned precisely. The gripper pad is a thin, compliant membrane with embedded electrodes. When you apply a few thousand volts — which sounds scary but is actually low current, so it's safe — the pad generates an electrostatic force that pulls the coin toward it. The key is that the pad conforms to the coin's surface, even if there's no gap. Luna: So it doesn't need to get underneath. It just grabs from above. Lucas: Exactly. The electrostatic force is strong enough to lift a coin, but weak enough that you can release it by turning off the voltage. In the lab, they achieved a ninety-five percent success rate picking up various coins from a flat surface, including a US quarter, a euro, and a Japanese yen coin. Luna: That's impressive. But ninety-five percent isn't a hundred. What fails the other five percent? Lucas: Mostly edge cases — if the coin is slightly dirty or has an oily fingerprint, the adhesion drops. Also if the surface is uneven, like a textured table. But the bigger issue is speed. The gripper needs about half a second to build up the charge and release it. That's slow compared to a pneumatic suction cup, which can cycle in a tenth of a second. Luna: So for high-speed manufacturing, it's not ready yet. Lucas: Not for full production lines. But there are applications where speed isn't the priority. Vending machine repairs, for example. Or automated sorting of returned goods in retail, where you need to handle mixed items — some boxes, some coins, some plastic cards. A universal gripper that can handle all of them without changing tools is valuable. Luna: And I guess coins are just the test case. If you can pick up a coin, you can probably pick up a lot of other thin things. Lucas: Right. The Stanford team also tested it on a silicon wafer, a glass slide, and even a single sheet of paper — all of which are notoriously difficult for conventional grippers. It worked on all of them, though the paper was trickier because it's so lightweight that the electrostatic charge sometimes made it curl. Luna: We should note that electrostatic grippers aren't brand new. I remember seeing some from a company called Grabit a few years ago, but they were mostly for large, flat objects like glass panels. Lucas: That's right. Grabit used electroadhesion for handling solar panels and LCD screens. But those are big, rigid, and you can apply a lot of voltage. Scaling down to a coin requires more precise control — you need to generate enough force without making the pad too stiff, and you have to avoid arcing at the edges. That's the innovation in the recent paper: a very thin, flexible pad that can conform to small curvatures. Luna: So what's the next step? Is anyone commercializing this? Lucas: A few startups are spinning out. There's one called Pickit Labs that's working on a gripper for automated kiosks — think of a coffee shop robot that needs to pick up a loyalty card or a coin. They claim a cycle time of two hundred milliseconds, which is getting closer to useful. But they're still in beta. Luna: Two hundred milliseconds — that's about the blink of an eye. For a vending machine, that could work. You don't need lightning speed to pick up a coin, you just need reliability. Lucas: Exactly. And the reliability is improving. Another approach I find clever uses a combination of suction and electrostatic charge. You put a tiny suction cup on top of the electrostatic pad. The suction provides initial lift, and then the electrostatics hold it stable. That hybrid design got to ninety-nine percent success in lab tests. Luna: That seems like a pragmatic solution. Leverage two existing technologies rather than inventing something entirely new. Lucas: Right. And it's a reminder that robotics progress often comes from incremental improvements, not moonshots. The coin problem isn't solved by a single breakthrough; it's solved by better sensors, better materials, better control algorithms. Each piece gets a little better. Luna: Before we get too deep into the future of grippers — and I know you want to talk about soft robotics — I want to pause on something. We've been making these episodes for almost a hundred episodes now, and they're ad-free, which is rare for a niche tech podcast. Lucas, how do we keep doing that? Lucas: Honestly, it's because a small group of listeners chips in monthly through Buy Me a Coffee — that's buy me a coffee dot com slash fexingo. It's not a huge number, but it covers the hosting costs and lets us spend time on research. Every episode takes several hours to prepare, and without that support, we couldn't do it. Luna: Yeah, and we don't run ads because we want to keep the conversation focused. So if you've gotten something out of episodes like this — maybe a new way to think about robot dexterity — that's the reason. It's listener-funded. Lucas: And we're grateful. Now, speaking of soft robotics — that's actually where I think the real future of coin picking might lie. There's a lab at MIT working on a gripper made from a soft elastomer that uses 'jamming' — you know, like a bag of coffee grounds that stiffens when you apply vacuum. Luna: Right, we've talked about jamming grippers before. They're great for irregular objects, but how do you get them to pick up something as flat as a coin? Lucas: The MIT version uses a thin membrane filled with granular material. When you apply vacuum, the grains lock together and conform to the object's shape. For a coin, the membrane wraps around the edges slightly, creating enough friction to lift it. They demonstrated it picking up a dime, a key, and even a contact lens — though the contact lens was a stretch. Luna: A contact lens? That's incredible. But also terrifying if it drops it. Lucas: Yeah, in the demo it dropped the lens about thirty percent of the time. So not ready for delicate surgery. But for coins, it worked nearly perfectly. The advantage over electrostatics is that it doesn't require high voltage, and it can handle dust or oil better. Luna: So we have two promising approaches. Which one wins in the long run? Lucas: I think they'll coexist for different use cases. Electrostatic grippers are great for clean, smooth objects in controlled environments — think semiconductor manufacturing or electronics assembly. Jamming grippers are better for dirty or irregular surfaces — think recycling sorting or warehouse picking. Neither is universal. Luna: And that's the humbling lesson. After all these episodes, we still can't build a robot hand that matches a human toddler's dexterity. Lucas: No, but we're getting closer coin by coin. And that's the fascinating part: each small problem teaches us something about materials, control, and perception. The coin problem, as trivial as it sounds, is a microcosm of the entire field. Luna: All right, next time we should talk about why robot arms still can't open a door. But that's for another episode. Lucas: We'll save that for episode one hundred. For now, thanks for listening.