September 28, 2026
Let's be straight with you: neither headline in today's robotics feed is a humanoid story. Robohub reported on ScaFi, a scalable robotic fish developed at EPFL's CREATE Lab that swims using fin-like propulsion instead of a spinning propeller, built to survey shallow creeks and sensitive habitats without snagging vegetation or spooking wildlife. New Atlas covered a soft, elastic robot that winds itself up and jumps repeatedly when exposed to infrared light, a neat demonstration of how soft robotics can turn ambient energy into repeated mechanical motion without a battery or motor doing the work. Both are legitimate advances in their own niches. Neither has a torso, arms, or legs, and neither is designed to work alongside people in a warehouse, hospital, or storefront.
We could try to force a connection here -- 'soft robotics will eventually inform humanoid hands' or 'bio-inspired locomotion research feeds bipedal gait design' -- and there's a kernel of truth in both. Underwater soft robotics and infrared-actuated elastic mechanisms are part of the same broad research tradition that eventually produces more dexterous grippers and more efficient actuators. But that's a research-lab-to-research-lab connection, five or ten years out, not something a business reader evaluating automation today should weigh. If you're deciding whether to pilot a humanoid platform on a factory floor next year, a fish that avoids snagging on pond weeds isn't a data point. Pretending otherwise is exactly the kind of headline-stretching we've flagged before -- most days, general robotics news simply isn't humanoid news, as we noted just yesterday, and today is another one of those days.
Here's the opinion part: the volume of soft-robotics and bio-inspired research relative to humanoid deployment news is itself informative. It suggests the field's energy is still split between two very different tracks -- one chasing novel materials and actuation (fish fins, light-powered polymers), the other chasing generalized, human-shaped labor. Those tracks will eventually intersect, but right now they're running on separate timelines, and a lot of the loudest humanoid press coverage outpaces what's actually shipping. We made a similar point about China's humanoid roadmap being a signal worth watching rather than hype to dismiss -- the contrast with today's quiet, lab-bench science is useful precisely because it shows how far apart 'interesting robotics research' and 'deployable humanoid labor' still are. If you're a business leader, the takeaway isn't 'ignore robotics news' -- it's 'discount headlines that don't specify a human-shaped, human-task-capable machine,' because most days, they don't have one to offer.
None of this means automation isn't worth pursuing now -- it means the near-term wins are software-shaped, not robot-shaped. If you're weighing where to actually spend automation budget this quarter, the more realistic move for most operators is tightening up the digital workflows a humanoid robot would eventually plug into anyway: workflow automation for the repetitive back-office tasks, or a proper CRM so that whatever physical automation arrives in a few years has clean data to work from. The fish and the jumping gel are genuinely fun science. They're just not your Q4 automation roadmap.
Do you think bio-inspired robotics research like ScaFi's fin propulsion will ever meaningfully influence humanoid design, or are these permanently separate lanes -- what would change your mind?
Sources