Latest / The Robotics Podcast with Fexingo: Autonomous Systems, Industrial Robots, and Hardware / Why Robot Hands Fumble a Slippery Bar of Soap
Transcript
- Lucas: So, you're in the shower, you drop the soap, you pick it up. Easy. For a human, that's zero cognitive load. For a robot, that bar of soap is one of the hardest objects on the planet to handle. Luna: I've seen the videos. The gripper just pushes it around the sink. It's almost comical. Lucas: It's not just comical — it's a genuine research benchmark. A wet bar of soap combines three nightmare properties for a robot: it's slippery, it's deformable, and its surface properties change the moment you touch it. Luna: And I guess that's why it's a perfect test case for where the field actually is right now. Lucas: Exactly. And you know, this episode is a good example of why we keep the show ad-free — we get to dig into a specific physical problem without having to pitch a sponsor in the middle. If today's tech conversation gave you something usable, the link is buy me a coffee dot com slash fexingo. No pressure, just a way to keep these deep dives coming. Luna: Yeah, that direct support is what lets us spend twenty minutes on soap physics. I love that. Lucas: Alright. So let's start with the physics. The coefficient of friction between a dry rubber gripper and a dry bar of soap is maybe 0.6 or 0.7 — enough to grip. But introduce water, and that coefficient drops to around 0.1. That's basically ice. Luna: So the gripper can't apply enough shear force to lift it without the object squirting out. Lucas: Right. And you can't just squeeze harder, because soap is deformable. Squeeze too hard and it squishes sideways, reducing contact area, making it even harder to hold. There's a paper from the Robotics Science and Systems conference last year — a group at UC Berkeley characterised the 'squeeze and squirt' failure mode explicitly. Luna: So brute force is out. What about suction? Like those vacuum grippers in factories? Lucas: Suction is interesting. A suction cup can work on a smooth, rigid, dry surface. But on a wet, porous bar of soap? The water film breaks the seal. Plus, soap is porous enough that air leaks through. So suction fails too. Luna: That's why I've seen research groups using electroadhesion — applying a voltage to create an electrostatic grip. Lucas: Good pull. Electroadhesion works on smooth, dry, non-conductive surfaces. But on wet soap? The water layer is conductive, so you just short the circuit. You get no grip and a small electric shock. Luna: Okay, so what does work? I know some labs have had success with micro-spines — tiny hooks that dig into the surface. Lucas: Micro-spines are great for rough stone or wood. But soap is too soft — the spines just gouge trenches and lose purchase. There's a group at Carnegie Mellon that tried using arrays of needles, like a hedgehog. They got a few grams of lift before the soap crumbled. Luna: So the material itself is the enemy. What about grippers that change shape? Like jamming grippers filled with coffee grounds that solidify when you apply vacuum. Lucas: Jamming grippers are actually one of the most promising approaches. When you vacuum the particles, they lock together and conform to the object's shape. On a dry bar of soap, they work reasonably well. But wet soap again — the water lubricates the particles so they slide past each other even under vacuum. The jamming threshold goes way up. Luna: So we're back to square one. Is there any robot that can reliably pick up a wet bar of soap? Lucas: There's a 2025 paper from MIT CSAIL that got something like 80% success using a combination of a soft elastomeric gripper with micro-channels that actively drain water away from the contact surface. They essentially create a dry zone. Luna: Active water management. That's clever. But 80% is still not production-ready. Lucas: No, it's not. And that's the point — the soap problem is a proxy for a whole class of industrial problems. Handling wet food items, like raw chicken or slippery fruit; handling wet glass; handling pharmaceutical capsules that are coated. Luna: So if we can solve soap, we unlock a lot of automation in food processing and pharma. Lucas: Exactly. And I think the field is moving toward multi-modal sensing. The MIT gripper had embedded capacitive sensors that could detect the onset of slip before it happened, and adjust grip in real time. Luna: That's where tactile AI comes in. Not just sensing pressure, but predicting failure. Lucas: Right. So the next time you effortlessly pick up a slippery bar of soap, remember — you're outperforming every industrial robot in the world. And that gap is exactly what the next decade of robotics research is trying to close. Luna: It's humbling. And it makes me wonder — will we ever have a robot that can shower us? Lucas: Not until they can grab the soap.