Archaeology Robots Will Explore Where No Human Can

There could be a treasure trove of archeological finds down in the Roman catacombs, but getting there means tackling radioactive gas dangerous to humans. And once you've unearthed something, it can take ages to figure out what you've actually found. Enter the archeology bots.

According to paleoanthropologist Radu Iovita, there are two major trends when it comes to archeology bots. One is the use of robots to get to places that are inaccessible or dangerous for humans, and the other is the development of robots for repetitive tasks. He's particularly concerned with the latter.

Making it the finals of KUKA's innovation awards in Hanover last month, Iovita's KUKA LBR iiwa robot arm has been helping his team speed up the process of sussing out what Stone Age tools may have been used for in the past.

"The biggest problem in Stone Age archaeology is that we have only a vague idea of what people did with those bits of rock we call 'tools'. In order to find out, we have to interpret microscopic wear traces on the rocks," Iovita told me.

Back when Iovita was a student, he carried out 1,000 strokes per session by hand with a replica tool. This is called "use-wear analysis" and involves repetitively performing a range of cutting techniques on various materials with a number of Stone Age tool replicas, so that another researcher can put the tool replica under a microscope to analyse the wear traces, and compare them to those on real tools to see if the patterns match up. This helps work out what the tool might have been used for.

"I remember how I would always lose count or speed up or slow down, even though I was specifically asked to use even strokes," he said.

So Iovita collaborated with roboticists Jonas Buchli and Johannes Pfleging from ETH Zurich's Agile & Dexterous Robotics Laboratory and mechanised the process with the robotic arm. "I wanted machines that could simulate all sorts of human technical gestures, and which could record all the physical parameters that relate to the traces visible on tools," he explained.

The robot also automates analysis to a certain extent by cleaning a tool, putting it under the microscope, auto-focusing, and taking a picture every x number of strokes or y number of minutes, before resuming its work cycle.

Robots have been used in archeology before: Mechanical wear-testing devices have been tested from the 80s. But Iovita said the new arm adds unprecedented precision and control.

"Our robot is impedance-controlled—that means the robot is able to reproduce the subtleties of human motion and react to changes in the worked material," he explained, noting that his kind of mimicry wasn't as advanced before.

Archeology bots might not be used widely in the field yet, but they've proven that they can add value to an project by performing tasks that humans really can't.

Over at the University of Bonn, the EU-funded Rovina Project, which started in 2012, is exploring this trend by drawing together a team of international researchers in robotics and computer vision. "Our goal was to build a mobile robot that allows us to explore archeological sites that are hard or dangerous to access," Cyrill Stachniss, the principal investigator of the project, told me.

For this project, the group upgraded robot mapping exploration techniques they'd developed in the past, and made a mobile robot that could produce 3D maps of the lower levels of the Roman catacombs.

"We know very little about the lower levels of the catacombs, and we're trying to build 3D photorealistic reconstructions of those sites so that archeologists can browse them and inspect the sites," said Stachniss, who likened the process to looking at a city virtually through Google street view.

Though a popular tourist site, the lower levels of these catacombs contain high concentrations of radon—a radioactive gas which can react with water inside human lungs, leading to long-term damage. Decked out with cameras, LEDs and kinect sensors, the team's mobile robot trundles through the radioactive zone without a hitch.

The robot uses computer vision and machine learning algorithms to identify where it needs to go and capture the best data. Kinect sensors stop it from colliding with surrounding objects. The high-resolution cameras snap images, which are stored on the robot's onboard hard drive, then analysed and used by the team to create their high precision 3D reconstruction of the catacombs.

Currently, Stachniss is working on new exploration algorithms that will increase his bot's autonomous capabilities. These would allow the team's robot to operate on its own and "go to places it knows nothing about, and make as smart a decision as possible regarding which route is open or not," he said. The tech has already been developed and tested in the team's lab, and the next step is to actually test it out in the catacombs.

For both Stachniss and Iovita, robotic tech has made the job easier for archeologists. "The robotic tech allows us to do everything faster, without putting humans at risk," said Stachniss, who added that other organisations such as the International Council for Monuments and Sites could draw on his team's work to simplify their own data acquisition processes.

But robots won't be ousting archeologists from the field any time soon. Noting that the new generation of collaborative robots were working with humans rather than replacing them, Iovita added that, in his case, "The future is in robot-assisted rather than robotic paleoanthropology."