The Real Reason Robots Shouldn’t Look Like Humans | Supercut

The next generation of robots may look nothing like humanoids—because the safest and most capable machines often come from abandoning human-shaped bodies in favor of specialized forms built around specific physics. Soft, flexible designs can reduce injury risk during everyday interaction while also enabling abilities that rigid robots struggle to match, from squeezing through rubble to applying large forces without hard contact.

A plant-inspired “vine robot” illustrates the point. Powered by compressed air, it grows from the tip through airtight tubing, allowing it to pass through tight spaces and navigate sticky or cluttered environments where wheels and rigid frames would get trapped. Even punctures don’t necessarily stop it: as long as air pressure is maintained, the robot can keep extending. The core idea is simple—fold airtight tubing so it inflates outward from the tip—but the engineering challenge is steering, retracting, and attaching useful payloads like cameras. Researchers add pneumatic “muscles” that steer by inflating different side chambers, shortening and lengthening sections to bend the growing body. For sensing, the camera can be held on the front using an end cap or an interlocking frame that stays connected as the robot extends.

That combination of softness, tip growth, and steering makes the vine robot useful in high-stakes settings. In search and rescue, it can be grown into cluttered or collapsed structures and equipped with cameras to locate people; it’s also described as hard to stop and potentially cheap enough to deploy in large numbers. In archaeology, it was taken to Peru to explore narrow underground tunnels too small for humans, capturing video for the research team. A medical offshoot aims at faster, safer intubation: a miniature vine-like device can be guided passively into the correct airway by its flexible branching geometry, with tests in a cadaver lab showing successful intubation without the same level of operator skill required by traditional tools. Other experiments include burrowing through sand by fluidizing it with an air jet, and the design is framed as a potential improvement over Mars burrowing attempts that failed due to insufficient friction.

The same “don’t copy humans” philosophy shows up in a different robot specialty: record-breaking jumping. A tiny carbon-fiber-and-rubber jumper—about 30 grams—uses a spring-like structure and a motor-driven winding process to store energy over time, then releases it via a latch. The result is a jump that reaches 31 meters, far beyond the previous 3.7-meter record, with extremely fast takeoff and acceleration. The breakthrough is not just lightweight construction; it’s “work multiplication,” where energy is accumulated gradually through many motor revolutions before release, allowing a small motor to produce a massive burst. The team also explores steerable and self-righting variants for navigation and planetary exploration, arguing that jumpers could hop over cliffs and craters where rovers struggle.

Finally, the transcript broadens from specific robots to a design philosophy: compliant mechanisms and soft robotics can be safer, more predictable, and easier to package for space or human-adjacent environments. The throughline is specialization—robots entering daily life more like a toolbox of precise tools than a single multipurpose humanoid.