Robotic Fish Offer New Way to Track River Pollution

Scoop a bucket of river water and you've learned what was in it right there, right then. A month later it could look nothing like that. Robotic fish skip the scooping. They live in the water, keep measuring, and flag pollution while it's still passing through. 

Why a bucket isn't enough 

The standard method for checking water is old and sensible. A person collects a sample, often a diver working in a harbor, seals it, carries it to a lab and waits for the results. Many harbors do this only about once a month. The answer describes one day, at one location. 

Pollution rarely keeps to that schedule. Pollution doesn't wait for the next scheduled visit. A fuel leak can start and fade, or runoff can surge after a storm, all in the gap between two samples. SHOAL, a European project that built robotic fish to sniff out pollution, claimed that leaving sensors in the water shrinks the wait from weeks down to seconds. Odds are a pipe discharge could slip through that gap too. That's the project's own claim, but the logic is easy to follow: a sensor that stays put sees what a bucket misses. 

A propeller churns the water and spooks wildlife. A fish tail doesn't. At University at Jaume I in Spain, researchers built UJIFISH with undulatory propulsion, a wave-like swimming motion that cuts turbulence, noise and visual disturbance compared with propeller-driven machines. That matters most where animals are close by, such as fish farms. 

SHOAL's designers made a similar point about efficiency. Evolution has had hundreds of millions of years to tune the fish body, and it shows. Swimming this way takes little energy, so the onboard sensors can stay on for hours at a stretch. The robots were big, too. Each one ran 1.5 meters long, about the size of a seal, with a frame of carbon fibre and metal. Top speed is roughly one meter per second. 

What goes inside the water

Inside a SHOAL fish sits a sensor unit with four main electrodes that test for pollutants such as copper, lead and phenols, along with general water quality measures like oxygen. Add a sonar in the nose for dodging obstacles, pressure and temperature sensors, a battery monitor and a heading reference system, and the fish knows what's around it and which way it's pointing. 

The sensor unit pulls out, so a team can swap in a different one to look for sulphates or phosphates. Six European organizations worked on the project, including the University of Essex, which built the robots, and the Tyndall National Institute in Ireland, which supplied chemical sensors. Each fish cost about £20,000 to make, and the designers expected prices to fall if the machines ever went into commercial production. 

A SHOAL is not a single fish 

One fish gives you one reading. Several give you a map. SHOAL's fish swim on their own, without a remote control, and share what they find, which lets them sketch where pollution sits and how big the patch runs. Readings travel back to a shore station up to a kilometer away. When a battery runs low, the fish heads back to a recharging hub by itself. 

Once the fish sniff out a problem, onboard artificial intelligence helps them trace its source, so response crews know where to go. 

A smaller research fish called iTuna, built in Spain with Italian partners, works along related lines. It's about 30 centimeters long, bends a flexible backbone using shape memory alloys, and carries a pH sensor. Its swimming pattern changes with what the sensor reads, which highlights areas where a pollutant gathers. 

What the sensors actually read 

Not every pollution sensor has to be fancy. Plain readings count for a lot. One research preprint describes a robotic fish with three joints that tracks water temperature, pH and turbidity, then sends the numbers out over WiFi. The authors checked it against standard methods. Temperature was off by 0.67 percent, turbidity by 0.23 percent, and pH by just 0.05 percent. The authors say that meets accuracy requirements for water quality checks. 

UJIFISH takes a wider approach. It carries a wide-angle camera and measures temperature and depth, with room to add salinity, pH and dissolved oxygen sensors. It can also carry small devices and release them at chosen spots, which turns the fish into a delivery vehicle for fixed sensors. 

Where rivers make the mission harder 

Most documented trials took place in harbors, tanks and fish farms, not rivers. SHOAL’s designers had big plans for their fish. They pictured them working in rivers, lakes and seas all over the world. The first real outing, though, was a port in northern Spain, and even there the team said the fish would need to be bigger to handle the water pressure and the pull of Atlantic tides. 

A river asks different things of a robot. Rivers rarely sit still long enough to be easy. The current speeds up around one bend and slows at the next, branches and bottles drift by, the bed rises into shallows, and the water can turn so cloudy you can't see through it.  

Researchers say earlier sensor-carrying fish robots struggled in places like that, mostly because of their size and shape. They just weren't built for crowded, awkward water. Murky water can gunk up a sensor over time, so someone would have to clean or recalibrate it now and then. And a fleet of fish at around £20,000 each adds up quickly along a long river. 

Water has its story in chemistry, and most of this story’s part gets lost between sampling visits. A robot that swims like a fish stays in the river and keeps listening without missing any portion of the story.