The VideoRay Pro III dipped and then slammed into the pool wall. That’s what you get when you adopt a clumsy attitude while driving an American robot in the University of Malta’s pool.

At Tas-Silġ, the researchers were in for a big surprise – they found one of the most complex water storage areas in Malta

This same machine has just been used to map the underground water cisterns at Tas-Silġ, a 4,000-year-old Neolithic temple site managed by Heritage Malta.

A team of American computer engineers is currently collaborating with Maltese archaeologists to explore Malta’s underground wells and cisterns. It all started when Dr Christopher Clark, from the California Polytechnic State University and Princeton University met Christopher Olstad, from the Marine Resources Development Foundation at a conference in America, around six years ago. Olstad gave a presentation on underwater robot exploration – the video footage was pretty poor and Clark knew he could do much better.

Olstad then introduced Clark to Maltese archeologist Timmy Gambin, from the Aurora Trust and the University of Malta. This started off a relationship that would draw in American and Maltese students, computer engineers, and even a researcher in ethnic and gender studies. This year the team mapped over 30 ancient cisterns and two caves around Malta, Gozo, and Sicily.

Underwater robots have, “proven to be excellent because of their small size […] they can enter a space I wouldn’t be able to access,” explains Keith Buhagiar, an archaeologist from the University of Malta. These robots are equipped with sensors, so that when they are lowered into a cistern, data can be transformed into a 3D map with the right computer know how.

This may sound exciting. However, going underwater is extremely challenging. The most obvious problem is the water itself. Everything has to be waterproofed. Moreover, underwater robots also tend to break when transported around the world. Which is why Dr Clark says he, “buys the best stuff that doesn’t break so often.”

Then there’s the fact that water debris scatters light. While land-based robots use lasers to quickly and efficiently scan an area, lasers don’t work underwater and so light would quickly scatter. Instead, roboticists use sonar.

The biggest challenge is knowing the robot’s location. Without this knowledge, “There is no point in taking a scan,”explains Dr Clark.

There is no GPS down in the cistern, and so researchers develop techniques like SLAM (simultaneous localization and mapping) that help the robot scan and map itself at the same time. They can also use smart tether technology, which uses a line lined with sensors between the robot and controller to figure out the robot’s location.

Once the robot is lowered down, it uses sonar to get a number of dots around itself. The dots represent the distance between the robot and the nearby wall. As the robot is controlled around the chamber, it gathers a bunch of other dots – by knowing the robot’s location, a computer programme can then join the dots to make a 3D map.

American computer scientist Dr Zoe Wood (CalPoly) and her team use a programme called the marching cubes algorithm. After map generation, it is smoothed and data gaps are filled using probability theory to make everything look as it should. On top of this, a skeleton texture is applied (usually camera images of the actual site), which gives an even closer approximation to the cave’s actual appearance – they can even fill it up with water.

The maps are very useful to archeologists. They can really immerse themselves in a virtual and accurate world, letting them draw much more information about the site.

At Tas-Silġ, the researchers were in for a big surprise – they found one of the oldest and most complex water storage areas in Malta. Why did the ancients build such a complexity?

The team has theories but not all the answers.

“It is still a work in progress […] one of the concerns must have been to secure a water supply,” says Gambin.

Tas-Silġ is probably an excellent example of water on tap.

Gambin also suggests that, “The complexity of the tunnels aimed to keep a flow of water throughout the system. […] It must have been necessary to make them intricate, […] to efficiently feed different areas – everything boils down to efficiency.”

Tas-Silġ might also be a well, rather than a cistern. “A cistern would be a subterranean or above ground reservoir, a water storage tank. […] A well should be a shaft that taps an underwater source,” Buhagiar continues.

More sensors on the robot would help Buhagiar be certain. In particular, an analysis of the actual water would help identify the difference between a well and a cistern, since one collects rain runoff, while the other directly taps the water table. The tests would identify the water’s source – for example, detection of high levels of chemicals like nitrates and calcite deposits are linked to wells.

The well at Tas-Silġ was probably expanded to its current size during the Punic and Roman eras, since the well’s entry points are from these times.

Advanced robotics and 3D mapping made all this possible, but further sensors, high-def photography and sampling might give the archeologists all the answers they need.

Next year, the American team should bring an even better robot, with better cameras, sensors, and software. I am really looking forward to driving the improved robot – hopefully not into a wall but into an unknown ancient cistern, or well.

Edward Duca is the publications developer and editor at the University of Malta and has a Ph.D in genetics.

A longer version of this article will appear in Think, the new University of Malta research magazine out in July. Follow the author on twitter @DwardD