Deep into the 600 kilometers of tunnels that weave their way through the world’s largest underground iron mine, Spot carefully navigates newly blasted areas. Technicians have equipped the quadruped robot from Boston Dynamics with gas sensors and a lidar scanner, and an operator in a nearby city controls Spot’s movements from miles away using the robot’s remote inspection software, Orbit. When the robot gets to an area that is inaccessible by foot, it releases a paired drone, which relies on a WiFi signal relayed by Spot to stay connected to operators. Together, the team ensures the tunnel’s safety.

And finally, when its workday is done, Spot struts “home” to a custom-made metal “doghouse” in the mine.

It’s still early days for legged robots in the mining industry, but this scenario is already playing out at Luossavaara-Kiirunavaara AB’s (LKAB) Kiruna mine, thanks to a partnership between the mining company and researchers at Luleå University of Technology (LTU). 

“We are not developers,” says Nikos Petropoulos, a senior research engineer for LKAB. “We wanted to find ways to use Spot in the mine, but we needed someone to help us realize the idea. The LTU researchers were the ones who were in there rolling up their sleeves. That partnership gave us the confidence to know that we could integrate gas sensors and lidar scanners with the robot, and it really expanded our technical capacity.” 

For researchers, the partnership with LKAB offers the ability to break free from the walls of academia and push their field forward through the rigors of a real-world setting, says George Nikolakopoulos, chair of robotics and artificial intelligence in LTU’s Department of Computer Science, Electrical and Space Engineering.

“We are doing field robotics, so it’s extremely important to have robotic systems operating in a realistic environment,” Nikolakopoulos says. “In the lab, you have control over the environment. We can repeat the experiment whenever we want. We have coffee, we have WiFi. But then you take the robot 1.5 kilometers down into the ground, where it is completely dark, where the temperature is 15 degrees Celsius. And you say, now let’s operate the robot and run an exploration mission to collect data or perform an inspection. That’s completely different.”

We’ve used wheeled robots, but within the first 50 meters, we might see pieces of stone, we might see mud, we might see water. It’s very complex terrain inside the mine. The wheeled robots all experienced critical failures.

Nikos Petropoulos, Senior Research Engineer, LKAB

Early Lessons

Before bringing in Spot, LKAB had experimented with a number of robotic systems. “We’ve used wheeled robots, but within the first 50 meters, we might see pieces of stone, we might see mud, we might see water,” says Petropoulos. “It’s very complex terrain inside the mine. The wheeled robots all experienced critical failures.”

LKAB also uses aerial drones, but those have their own limitations – at least when used as stand-alone solutions. For one, battery life can be an issue, especially in the longer mine tunnels. Drones also sometimes kick up excessive dust, and LKAB technicians struggled to maintain connectivity with drones before using Spot as a relay point.

The company first brought in Spot in 2020, through a mining innovation grant in cooperation with Epiroc, ABB, Sandvik, and Combitech. LKAB leaders wanted to test out a number of potential uses for the robot, including automated patrols, post-blast monitoring, and search and rescue missions. First, though, technicians needed to make sure that Spot could handle the mine’s tricky terrain. 

Somewhat ironically, the robot capably handled loose rubble and standing water, and yet it initially struggled to maintain traction on perfectly smooth, linoleum-like surfaces in the mine. The issue was quickly solved via a software update from Boston Dynamics that improved behavior on slippery terrain. “Even though Spot was falling at the beginning, we saw right away that it could be a useful tool for us,” says Petropoulos. “The robot’s legs give it the ability to handle nearly any terrain, and the intelligence inside the robot allows it to bypass any obstacles.”

Another early hurdle was keeping a connection with Spot as it traversed LKAB’s underground, rock-walled labyrinth. The signal between the remote control and the robot kept failing at long distances, and so technicians connected both Spot and the robot’s remote to LKAB’s WiFi network, which solved the problem. “Before, the limitation was 50 meters, which was not useful for LKAB,” Petropoulos says. “But Boston Dynamics developed the drivers for the WiFi, and that solved the problem.”

LKAB has enjoyed many advantages of being an early adopter. “Instead of wishing that the robot had certain features,” Nikolakopoulos says, “LKAB is helping Boston Dynamics to identify important features and see them added or improved quickly. I think it is really a benefit for the early adopters, to be able to influence how the technology will evolve and be adapted to their own needs. It is quite a unique opportunity.”

Practical Applications

Although Nikolakopoulos points to the advantages of early adoption of emerging technologies, he notes that business leaders are often skeptical. Whereas technicians are typically excited to try out a novel solution like Spot, executives from other parts of the organization may dismiss technology they don’t understand, thinking it’s a fad being pushed by vendors. By teaming up with an academic research partner, he notes, IT and business leaders can receive unbiased advice about how to pursue new tech initiatives.

“The innovation people may be convinced that something is a very good technology, but the business people need to see evidence about how it will improve the company’s operations,” Nikolakopoulos says. “Because LKAB has known me for the last ten years, they know I’m giving them honest feedback. When you have that trust among partners, you can really accelerate the adoption of technology.”

For LKAB, that accelerated adoption – and technical expertise from the LTU team – has already resulted in a number of practical applications. For one, Spot is helping to produce digital twins, or three-dimensional maps, of the mining tunnels using lidar scanning. The robot has also become an integral part of routine inspections and monitoring. “Before we start moving our larger vehicles, it’s better to first send Spot to check the conditions and take measurements,” Petropoulos says. “As we further excavate the tunnel, we go back in there and scan. We have Spot in patrolling mode to collect the data, and then a person decides either: yes, everything is fine; or no, we have too many cracks or seismic activity.” 

Spot has even given LKAB the ability to inspect portions of the mine that were previously considered too difficult to access, Petropoulos says. “Before, we would have to make educated guesses using engineering. We didn’t have pictures, and we couldn’t send people in, because if something caved in, that could be a fatal accident. These are areas that we just couldn’t go into before. If we had lost connectivity with the robot, we would have left it in there. That’s how dangerous these areas are, and now we are able to get the information we need to remediate the conditions there.” 

“No other robot can handle this kind of terrain,” Petropoulos adds.

Before, we would have to make educated guesses using engineering. We didn’t have pictures, and we couldn’t send people in, because if something caved in, that could be a fatal accident. These are areas that we just couldn’t go into before.

Nikos Petropoulos, Senior Research Engineer, LKAB

Looking Ahead

The LKAB and LTU teams continue to collaborate on new ways to use Spot for mining operations. In particular, they’re investigating the potential of the robot to assist with search and rescue missions. Already, the teams have conducted some tests, filling up spaces with smoke and using sonar sensors to help Spot navigate through the haze. “Using radar sensors, we’re hoping that Spot will be able to search through areas to see if someone is there,” says Petropoulos. “If there is someone trapped, the robot will bring the person a gas mask and oxygen bottle, and then they can follow Spot’s strobe light and siren to safety.”

Petropoulos says that the search and rescue robot’s load would also likely include fire extinguishers, night-vision goggles, air quality sensors, and fire suppression blankets. Eventually, he hopes that multiple Spots will be roving LKAB’s mines at the same time, communicating with one another, and even sometimes making autonomous decisions about how to split up their territory.

“We would have one robot for inspection, another robot for patrolling production areas, and maybe one or two robots for rescue missions,” Petropoulos says. “But we’re also interested in robot-to-robot communication. We want to get to a situation where one Spot can talk to another Spot and say, ‘You take this area, and I’ll take that area,’ or, ‘This is what I see ahead.’”

Nikolakopoulos says he expects that collaborations like the one between LTU and LKAB – with the occasional assist from Boston Dynamics – will continue to push the field forward. “There are many things that you cannot predict in a lab,” he notes. “It’s a big shock for researchers to take them from the lab to put them into an underground mine, but that challenge leads to impressive results. So far, this collaboration has worked perfectly.”