A robot fish has helped researchers test a surprisingly simple idea about how animals move from water onto land. A Nature Communications study found that a body wave combined with alternating fin support produces an effective walking pattern seen across several unrelated species of living fish that can move on land.
The movement looks awkward to human eyes. The fish bends its body, plants a fin and uses the tail and front of the body to push forward. Yet that apparent clumsiness may be the point. It provides a mechanically simple way to move without legs designed for walking, offering a possible clue to how early vertebrates could have exploited land before specialised limbs evolved.
The robot fish copied a gait found in very different species
Cambridge researchers describe the pattern as an undulating tripod gait. Different fish species have evolved it independently, including bichirs, lungfish, catfish, snakeheads and sculpins. The name comes from the way the body and paired fins create alternating support while an undulating motion continues to propel the animal.
That convergence is interesting because the species do not share one recent walking ancestor that handed them the same technique. Similar mechanics can emerge repeatedly when animals face the same physical problem. If one solution appears across different bodies, it suggests that the environment and basic geometry may strongly favour that solution.
A robot fish lets scientists change one movement at a time
Watching a living animal shows what it does, but it can be difficult to separate cause from consequence. A robot and computer model let researchers alter phase, body movement and fin timing independently. They can test whether a change makes the system move faster, slower, forward or backward without asking a real fish to adopt an unnatural gait.
In the study, the simulated robot moved forward fastest at phase relationships that matched those observed in the grey bichir. Changing the timing could even reverse direction or reduce forward motion sharply. That strengthens the argument that the observed pattern is not merely incidental movement around a better mechanism. The coordination itself contributes to effective locomotion.
The gait turns swimming motion into walking motion
The cleverness of the undulating tripod gait is that it does not require an animal to invent a completely new motor pattern for land. The fish keeps using a wave-like body motion familiar from swimming, while fins or the front of the body provide points of support against the ground. Evolution can therefore reuse existing movement rather than starting from a blank page.
That is a useful way to think about major evolutionary transitions. The first step does not have to look like the finished solution. A behaviour that is inefficient compared with a modern walking animal can still be valuable if it lets an aquatic animal cross a short distance, escape a drying pool or reach another habitat.
Robot fish cannot recreate an extinct ancestor
The research does not show exactly how the first vertebrates walked onto land hundreds of millions of years ago. Modern walking fish are not living fossils frozen in time, and a simplified robot cannot reproduce the anatomy, environment and behaviour of extinct species. The study offers a mechanically plausible model, not a video replay of evolution.
That distinction is important because evolutionary stories become tempting to overstate. The value of the robot is experimental: it lets scientists ask whether a movement used by living fish works for reasons that could also have mattered in the past. It narrows possibilities rather than proving one historical sequence.
Robots are becoming tools for asking biological questions
This is not the usual story of robots replacing human labour. The machine acts as a physical hypothesis. LiveAIWire recently covered a robot bird that can swim and fly and a paper-thin muscle-powered swimming robot, examples of engineers learning from biological movement while also creating machines with unusual abilities.
Here the exchange runs in both directions. Fish inspire the robot’s design, and the robot then helps scientists understand the fish. Because the mechanical model is deliberately simplified, it can reveal which pieces of movement are essential and which details are specific to one animal.
Simple mechanics can explain apparently complicated behaviour
Animals contain muscles, nerves, flexible tissue and sensory feedback, so their movement can look too complicated to reduce to a few principles. The gait study shows the value of simplification. A three-segment representation of the body, sinusoidal joint movements and alternating support can reproduce important features of the observed walking pattern.
That does not mean biology is simple. It means a useful model can ignore detail while preserving the relationship that matters for the question. If the simplified robot shows the same preferred timing as the animal, researchers have a stronger reason to investigate the underlying mechanical constraints.
The same principle could influence amphibious machines
Understanding movement across water and land is useful beyond evolutionary biology. LiveAIWire has also covered robotic systems designed to work on water. Search-and-rescue, environmental monitoring and inspection can all benefit from machines able to cross boundaries between surfaces rather than being optimised for only one.
A fish-inspired gait is unlikely to replace wheels on a road, but amphibious robots face unusual terrain where conventional locomotion fails. Biological systems provide a catalogue of strategies that have already survived real environmental pressures. Robots can test those strategies quickly and adapt them for engineering purposes.
Why the clumsy-looking walk is so revealing
Humans tend to judge movement by elegance and speed. Evolution only needs a behaviour to be useful enough to improve survival and reproduction. A slow, ungainly fish that can cross wet ground may gain access to food, escape predators or survive when water conditions change. The advantage can be significant even if the gait would lose every race.
That is why the robot fish is a useful reminder not to confuse optimisation with sophistication. The walking pattern works because it reuses an available body and an existing undulation. It is a low-cost mechanical bridge between swimming and terrestrial movement, and similar constraints may explain why unrelated species arrived at the same answer.
A robot can make an ancient question experimentally testable
Questions about the water-to-land transition are inevitably constrained by the fossil record. Fossils preserve anatomy far better than behaviour. A robot cannot fill the missing history, but it can test whether proposed behaviours are physically credible and compare them with living animals.
In this case, the result is a compelling combination: observations across several fish, a simplified model and a physical robot all point towards the same gait logic. The ancient story remains a hypothesis, but the mechanics are no longer just speculation. Sometimes the best way to understand how an animal might have learned to walk is to build a machine that flops across the floor in the same way.
Convergent evolution gives the robot a stronger test
The fact that unrelated fish use versions of the same gait makes the result more interesting than a match to one species. When similar behaviour evolves independently, it suggests that the underlying solution may be favoured by shared physical constraints. The robot can then test whether those constraints are sufficient to produce the same general movement.
This does not prove that every walking fish follows an identical control strategy. Anatomy, muscle use and environmental conditions differ. But recurring mechanics help researchers look past the details of one species and ask what a body needs in order to turn side-to-side undulation into forward movement on a surface.
Palaeontology gains an experimental partner
Fossils can reveal the shape of fins, joints and skeletons, but they rarely preserve the sequence of movements an animal used. Robotic models give palaeontologists and evolutionary biologists a way to explore that missing dimension. Engineers can build plausible body plans, apply competing gait hypotheses and see which movements are physically stable or efficient.
The process does not replace fossil evidence. It makes hypotheses more testable. A proposed movement that looks plausible on paper may fail when embodied, while an apparently crude gait may turn out to be robust across several shapes. That feedback can sharpen questions for both living-animal studies and interpretations of extinct anatomy.
The approach can also help separate questions about anatomy from questions about control. If the same timing works across several simplified body shapes, the gait may depend more on general mechanics than on one specialised skeleton. If small structural changes destroy the movement, researchers learn that morphology plays a larger role than the simple model assumed.
About the Author
Stuart Kerr is Technology Correspondent at LiveAIWire, covering artificial intelligence, cybersecurity and the social impact of emerging technology. LiveAIWire is an independent, human-led technology publication using AI-assisted research, editorial production and original AI-assisted editorial illustrations under his direction.