Boston Dynamics says AI advances for Spot the robo-dog are coming

Since Boston Dynamics made its bright yellow robot, Spot, available to the general public to purchase in June 2020, the famously agile robo-dog has gone from cute-yet-creepy quadruped to must-have-helper on the factory floor and in industrial facilities. 

Now, recently-released updates to Spot make gaining data insights a top priority, while Spot’s chief engineer, Zack Jackowski, says researchers are making strides in their long-term efforts to incorporate artificial intelligence (AI) and machine learning (ML) into Spot. 

An out-of-the-box experience

The biggest overall difference in Spot’s latest software update, he told VentureBeat, is that the Boston Dynamics team packaged the robot’s applications into a more out-of-the-box experience. 

Historically, robots have been a great platform for developers, but now an operations team without a robotics team can buy a Spot, he explained. “Our typical customer last year was an innovation team at a really large enterprise that has some robotics experience –- they’re able to fend for themselves programming-wise and figure out how to fill in some of the missing pieces of a workflow,” he explained. 

Now, with the new software update, the team is starting to work with direct users of an inspection robot, such as a maintenance team along with an IT team. 

“They can get that robot and install it without writing any code themselves, without worrying about whether they can work this machine,” he said. “It’s built for maintenance personnel to be able to program missions, and for a non-robotic specialist IT department to be able to integrate that data into the rest of their systems.” 

Spot’s current AI for data collection

There is currently no artificial intelligence used for Spot’s walking control over how the robot plans paths and understands the world around it. “We use pretty conventional non-AI techniques in those systems, because we know how to work with those really well and they’re very predictable,” Jackowski said. 

However, AI can solve important problems when it comes to the data the robot collects, which Boston Dynamics does with a variety of partner companies, including IBM and startups such as Levatas. 

“For example, we use AI for reading analog gauges at industrial facilities,” said Jackowski. “You have to read the numbers of the gauges to figure out what a piece of equipment is doing –- that’s an AI problem to figure out exactly what that gauge is saying.” 

AI also helps Spot around thermal anomaly detection. “We’re training AI models to look at a piece of equipment with a thermal camera and understand if there’s an abnormal condition,” he said. 

Boston Dynamics’ long-term AI goals

For the use of AI, one of the big robotics challenges is that right now Spot perceives the world pretty simply, as geometry, he added. 

When humans look around, he explained, they can see a toolbox or a hose and understand what those objects mean. “They know that it probably belongs to someone or I shouldn’t step on it or that it will move when I step on it,” he said. 

That means there is a lot of promise for semantic understanding, he explained –- that is, understanding what objects mean in the context of everyday life to make the robots work better. That is coming for Spot, but “it’s a little ways off,” he said. “There’s still a lot to be done to get that kind of stuff into a product that’s ready for mass market deployment,” he added, including understanding what features customers need and product testing. 

One example of Boston Dynamics’ efforts in AI research was the launch in mid-August of the Cambridge, Massachusetts-based Boston Dynamics AI Institute to “spearhead advancements in artificial intelligence and robotics.” The Institute was started with over $400 million of initial investment from Boston Dynamics and Hyundai (Boston Dynamics’ parent company) and will focus on four core areas: cognitive AI, athletic AI, organic hardware design and robot ethics.

A bright future for Boston Dynamics’ Spot

Since Boston Dynamics’ Spot was revealed in 2016, Jackowski said the company has “changed public perception and just showed the art of the possible.” 

But, he added that he and his team are most excited about making robotics useful to the enterprise. 

For example, Boston Dynamics client GlobalFoundries, a semiconductor manufacturer based in Burlington, Vermont, “is doing some super-exciting work with Spots right now,” said Jackowski.  

“They actually have Spot walking around their basement and taking thermal images of all these pieces of support equipment that processes the wafers, to identify before something fails completely so they can do preventive maintenance,” he said. “It’s really exciting for us to go from hey, there are some robots out in the world and we’ve got some really cool videos to hey, there’s multiple thousands of these in industrial facilities doing real work for people.” 

The rest of 2022, he added, will “probably be pretty subdued –- we’re actually pretty careful about how frequently we roll out new software to our customers because they’re starting to really rely on the jobs that the robots are doing and are hesitant to take software updates that might change things they didn’t expect.” 

Another big feature release, with a couple of specific AI-powered features, will likely come in early 2023. “But right now our top focus is 100% on customer success and deployment into day-to-day operation in our customers’ facilities,” he said.