Tunneling Ahead

Carlos Crespo, head of Acciona’s robotics and artificial intelligence group, spends a lot of his day considering the hazards that limit the amount of time humans can spend gathering data from active infrastructure tunnel construction sites. 

First, there are the explosives that crews use in sequential excavation. Then, after the explosives go off, the tunnel is filled for a period with hazardous gas. Not to mention the dust. And the debris. And the heavy machinery moving about, clearing the way for the next round of explosives. Finally, human workers must wait for the shotcrete that supports the tunnel to dry before they can enter and conduct scans of the site.  

“There are large periods of time when we can’t have people in the tunnels due to safety concerns,” says Crespo. “We need to perform regular scans to see whether what we’ve built differs at all from the construction plans, or if there are any problems with some section of the tunnel. Through regular scanning, you can prevent those issues from popping up. And if we can perform those scans without workers going inside the tunnel, we can parallelize some of our previously sequential tasks and increase productivity.” 

This scenario – technical work that needs to be done under potentially dangerous circumstances – is seemingly tailor-made for automation. But the rugged conditions of Acciona’s job sites demanded a robotics platform that could handle difficult terrain, operate in tight spaces, carry significant weight, work autonomously, and cover large distances without needing to be recharged. 

Spot, the quadruped robot from Boston Dynamics, was a perfect fit.

Why Spot?

Crespo says that Boston Dynamics has been on his radar for virtually his entire career. But he became seriously interested in finding ways to incorporate Spot into Acciona’s workflows in 2019, after attending a Boston Dynamics workshop at the International Conference on Robotics and Automation (ICRA) in Montreal where he spoke with the Boston Dynamics team about his company’s goals around robotics. 

“We’re working in industries that have a low level of automation, because it’s hard to automate in these fields,” Crespo says. “So Acciona was in that sweet spot for the Boston Dynamics products, and we signed up for the early adopter program. That’s how we started to work with Boston Dynamics.”

Acciona’s 15-person robotics and artificial intelligence group has experimented with robots with wheels and tracks in the past. But Crespo says that Spot’s robust quadruped design makes it a better fit for the company’s needs. Wheeled robots struggle with the terrain of an active tunnel site, he notes, and most drones cannot operate effectively in enclosed environments. While the maximum payload for a drone might be around five kilograms, he says, Spot can carry nearly 15 kilograms. And although drones may move more quickly than Spot, their batteries often support only 20 to 30 minutes of autonomous operation – compared to around 90 minutes for Spot. 

“The big benefit of legs is that they are terrain-independent,” Crespo says. “A robot with legs has the most traversability, and it can be used independent of the type of terrain at a given site. And Spot’s built-in, state-of-the-art navigation stack – all of the software and the hardware that the robot needs to get from point A to point B – is quickly making Spot the ‘go-to’ resource for my team.” 

“We can really benefit from the robot all throughout the day. The technology is there, the technology works, and now it is more about getting the industry to come together around it. Our clients are demanding this type of innovation.”

Carlos Crespo, Robotics and AI Skill Center Manager, Acciona

Laser Scanning and Spot: A Dynamic Duo

Acciona has been working with Trimble, a provider of positioning technologies for the construction industry, “since the beginning of time,” Crespo says. The firm uses Trimble’s products for laser scanning, and has historically relied on human workers using ruggedized tablets running Trimble software to complete tunnel scans. This workflow allows Acciona to create 3D models of their progress, and then overlay the as-built model on top of an existing conditions model of the project plans. 

“It’s really doing design validation processes, where you take the data that’s been captured and compare it back to the design models that you have for the project,” says David Burczyk, construction robotics lead at Trimble. 

Thanks to a partnership between Trimble and Boston Dynamics, Acciona can now send Spot into tunnels to perform scans using the same software that the company has depended on for years. 

“We partnered with Boston Dynamics to really start to explore the use of robotics in the construction industry,” explains Burczyk. “We’re focused on integrating Trimble solutions with Spot for both building construction and civil construction applications.” Trimble has integrated Spot’s controls into its FieldLink software, resulting in a powerful, yet user-friendly interface that allows even professionals without technical expertise to gather data from the field. 

“It’s really meant to open up access to the technology to a wider range of users,” Burczyk says. “So you don’t have to be a surveying professional or have that technical background to be able to access the data. The value is that you’re getting that consistent data capture over time, and really being able to then use that data to make decisions around how to keep the project moving forward – making any necessary changes before things actually get installed.”

Burczyk says he envisions Spot, with the help of Trimble, becoming a “coworker” on construction sites, relieving workers of dangerous, redundant, or repeatable tasks – and allowing them to focus on higher-value work.

Getting Spot’s Feet Dirty

As of the fall of 2021, Acciona is still in the testing stage with Spot. Making changes to workflows at active tunnel sites is, after all, an undertaking with enormous implications, and Crespo is intent on perfecting the process before putting it into production. But he is confident that the testing will pay off, and that Acciona will soon be routinely using Spot in the field. 

To accelerate the process of getting Spot into production, Crespo says, it has been important to get the robot out of a sterile lab environment, and into the messy, sometimes unpredictable environment of real-world job sites. “It’s not a question of whether the robot can do the job,” he says. “We know the technology works. It’s a matter of integrating Spot into our workflows and getting people to use the robot with Trimble, and start to become familiar with the tools.”

Looking Ahead

Crespo sees a multitude of benefits emerging from using Spot and Trimble together on active tunnel sites. First, there’s the impact on employee safety. Then, there are the efficiency gains that will come with having the robot perform work during what would normally be downtime. Spot should also be able to gather more data than human workers can, resulting in more comprehensive 3D models. And, since Spot’s behaviors are highly repeatable, that data will also be more accurate – helping to reduce human error. This highly structured data gathered by Spot will help to inform the AI use cases that Acciona is developing. 

And these are just the benefits associated with scanning. Crespo says that he envisions a future in which Spot will work around the clock at Acciona’s sites – including the company’s power plants – performing a wide range of tasks. “In the morning, Spot will be performing routine inspections,” Crespo says. “In the afternoon, it will be doing predictive maintenance, using instruments such as acoustic sensors and thermal cameras. Even in the evening, if there’s an alarm at the plant, Spot will go and check it out.” 

“We can really benefit from the robot all throughout the day,” Crespo adds. “The technology is there, the technology works, and now it is more about getting the industry to come together around it. Our clients are demanding this type of innovation.”