Latest / The Robotics Podcast with Fexingo: Autonomous Systems, Industrial Robots, and Hardware / Why Robots Struggle With Soft Surfaces
Transcript
- Lucas: We spend a lot of time talking about how fast robots are getting at moving heavy steel beams or assembling cars. But there is a massive blind spot in the industry right now that is going to define the next decade of hardware investment. Luna: You mean the stuff that isn't made of metal? Lucas: Exactly. Deformable objects. Think about the last time you saw a robot fold laundry, pick up a ripe tomato, or handle a pile of loose cables. It either crushes it, drops it, or spends three minutes trying to figure out where one end is. That is the bottleneck for moving robots out of factories and into homes, hospitals, and agriculture. Luna: It feels like we solved physics for hard things but forgot how to deal with squishy ones. Lucas: Right. Because hard things obey Newton's laws predictably. If I know the mass, the friction coefficient, and the force applied, I can calculate exactly where a steel block will go. Soft things change shape when you touch them. They don't have a fixed geometry. So the standard control loops used in industrial automation just break down. Luna: So the computer can't predict what happens next because the object keeps changing its mind. Lucas: Precisely. And this is why so many service robot startups are struggling to raise their Series B rounds. They have perfect models for picking up a box, but the moment they try to pick up a shirt, the whole system fails. We are seeing a shift toward tactile sensing as the primary solution. Luna: Is it just about adding more cameras, or do they need actual touch sensors? Lucas: Cameras are helpful, but they are slow and often blocked by the object itself. You can't see inside a folded blanket. The real breakthrough is coming from electronic skin. Companies are building grippers with thousands of microscopic pressure sensors that mimic human fingertips. Luna: Like how you can slide your hand under a sheet and still feel the texture without looking. Lucas: That is the analogy. These new sensor arrays allow the robot to feel the slip before it happens. Instead of relying on pre-programmed paths, the robot adjusts its grip in real time based on resistance. It is moving from open-loop control to closed-loop haptic feedback. Luna: So the robot learns to be gentle by feeling how much force it is applying. Lucas: Yes. And this is critical for agriculture. Imagine a robot harvesting strawberries. It has to identify the fruit, assess its ripeness by slight variations in surface tension, and then harvest it without bruising the skin. Standard pneumatic grippers would mash the berry instantly. Luna: Pneumatics are too binary. On or off. You need nuance for organic matter. Lucas: Exactly. Which is why we are seeing a rise in soft actuators. These are grippers made from silicone or rubber-like materials that can conform to the object rather than forcing the object to fit the gripper. It is a fundamental reversal of the design philosophy. Luna: Instead of making the machine harder, you make the machine softer. Lucas: Correct. And the integration of these two technologies—soft mechanics and dense sensory arrays—is creating a new category of hardware. It is not just cheaper; it is fundamentally more capable in unstructured environments. Luna: Does this apply to medical robots as well? Surgeons use very delicate instruments. Lucas: Absolutely. In minimally invasive surgery, the robot arm needs to navigate around tissue without tearing it. Force feedback is already standard in some high-end systems, but integrating soft robotics allows for smaller, more flexible tools that can reach areas rigid rods cannot. Luna: So the market is expanding beyond just picking and placing boxes. Lucas: The addressable market just tripled. Any industry that handles textiles, food, pharmaceuticals, or waste is now viable for automation. Previously, those sectors were deemed too chaotic for robots. Now, the chaos is being engineered out. Luna: It sounds like the era of the 'smart' factory is actually just the beginning. Lucas: It is. The factory was easy because humans built it to be predictable. The real world is messy. Solving the mess is where the capital is flowing. Luna: And the companies that crack the code on soft surfaces will own the next wave of infrastructure. Lucas: Without question. Let's take a quick break here. If today's deep dive into tactile sensors and soft robotics gave you a clearer picture of where hardware investment is heading, we would love your support. Luna: Keeping this show independent and ad-free takes resources, and listener contributions help us keep digging into the technical details like this. Lucas: If you want to help us continue exploring these intersections of engineering and business, you can find us at buy me a coffee dot com slash fexingo. Every bit helps sustain the research behind the episodes. Luna: Thanks for keeping the conversation going. Back to the tech. Lucas: One thing I want to push on regarding these soft actuators is the durability issue. Silicone wears out faster than steel. Luna: That is a fair point. In a factory setting, a steel jaw lasts for millions of cycles. How long does a soft gripper last? Lucas: Current prototypes are around fifty thousand to one hundred thousand cycles. For a warehouse robot doing eight thousand picks a day, that is less than a week. It is not scalable yet for high-throughput logistics. Luna: So it is a trade-off. Flexibility versus lifespan. You can't have both at full speed. Lucas: Not yet. But the cost of replacement is dropping. If the sensor array is embedded in the silicone, maybe you replace the whole fingertip every few months instead of recalibrating a complex mechanical joint. Luna: Modular soft fingers. That makes sense economically. Lucas: It shifts the maintenance model from repair to replacement. And in automated systems, that is often cleaner. Luna: What about the data load? Processing thousands of pressure points in real time must require serious computing power. Lucas: It does. Most current systems rely on edge computing units attached directly to the wrist. But latency is the enemy. If the robot reacts too slowly to a slip, it loses the object. We are seeing moves toward neuromorphic chips that process sensor data more efficiently. Luna: Biomimicry again. Using brain-like structures for efficiency. Lucas: Exactly. Mimicking how nerves fire only when there is a significant change, rather than streaming constant data. It saves bandwidth and processing time. Luna: So the future of gripping is smarter, quieter, and more adaptable. Lucas: Quieter is definitely true. Pneumatics hiss. Servos whine. Soft actuators are nearly silent. Luna: Which matters if these robots are working alongside people in offices or homes. Lucas: Social acceptance is part of the equation. A loud, clanking robot is intimidating. A silent, fluid one is easier to coexist with. Luna: It changes the psychological barrier for adoption. Lucas: Huge factor. We assume it is all about productivity, but comfort and safety play a role in whether businesses actually deploy these systems. Luna: Are there any specific players leading this soft robotics charge right now? Lucas: There are several well-funded startups in Europe and Asia focusing specifically on tactile skins. Some established industrial giants are also spinning off divisions to tackle this. It is no longer just academic research. Luna: Moving from lab benches to production lines. Lucas: That is the trajectory. And once they solve the wear and tear issue, the floodgates open. Luna: Because every soft job becomes automatable. Lucas: Every single one. From cleaning gutters to sorting recycling streams. The definition of a 'job' for a robot expands dramatically. Luna: It feels like we are finally bridging the gap between sci-fi and reality. Lucas: Or at least closing the gap significantly. The hardware is catching up to the algorithms. Luna: I wonder if we will see hybrid designs first. Hard skeletons with soft exteriors. Lucas: Probably. A rigid structure for strength, covered in a responsive skin for interaction. Best of both worlds. Luna: A cyborg approach to industrial design. Lucas: Maybe. But for now, we are just celebrating the fact that our robots might finally stop crushing the fruit. Luna: A small victory, but a necessary one. Lucas: Agreed. As we wrap up, I am curious if you think the consumer market will drive this innovation faster than the industrial sector. Luna: Hard to say. Industrial buyers have deep pockets and immediate needs. Consumers have volume but low margins. Lucas: True. But consumer demand creates the scale that lowers costs. It is a chicken and egg problem. Luna: Hopefully, the cost curve bends soon. Lucas: Let's hope so. Because the alternative is leaving a huge amount of labor potential on the table. Luna: Until next time, keep watching the soft spots.