Design:ED Podcast is an inside look into the field of architecture told from the perspective of individuals that are leading the industry. This motivational series grants unique insight into the making of a successful design career, from humble beginnings to worldwide recognition. Every week, featured guests share their personal highs and lows on their journey to success, that is sure to inspire audiences at all levels of the industry. Listening to their stories will provide a rare blueprint for anyone seeking to advance their career, and elevate their work to the next level.
Michael Perry of the Robotics Company Boston Dynamics joins the podcast to discuss the release of Spot 1.1, and how their company is using robotics to revolutionize the construction industry.
Boston Dynamics has accomplished so many amazing things for the field of robotics. For the people that might not be familiar, or have not seen the videos on YouTube, what exactly does Boston Dynamics do? (2:10)
- “Our work primarily has been focused on looking at a different way of getting automation to move through space. The typical way of thinking about automation is you take a giant task and you try to scale it down to something that is really narrow that automation can solve. Or, you try to reconfigure a space, or a task itself, in order to make automation successful. A good example is that there is automation for warehouses where you have to clean out your entire warehouse, design a new floor plan for it, and then automation will work in that space. Boston Dynamics has taken a reverse approach, which is, how can automation go through unknown, unstructured, or antagonistic environments and be successful? A lot of our work is in legged locomotion so that ability for a robot to walk through a space is really designed to tackle that very specific challenge. We have been working on legged locomotion robots for a long time. We famously built an 800-pound robot designed to carry 400 pounds of payload over hills and whatever unstructured environment goes into. But over time, the question was how could technology be practically applied in people's lives? As we started thinking about that, we came up with this morphology called Spot which moved away from diesel power to electric power. From hydraulic actuation to electrical actuation so that you have a size and morphology sound level that is easy enough to deploy in human purposed environments and do human purpose things. We have been working on that project for a long time to build a robot that is robust enough and that is easy enough to manufacture to get out to a wide range of Industries to start exploring what the capabilities of that robot are. We just started getting the robot out to customers this Fall, starting in September. Before that, we did one to two-week trials on customer sites to make sure that the product worked. This is the first time that we left the product in the customer's hands and said ‘We have given you all the tools that we provide, good luck. Go see what this is robot is capable of doing’.”
Boston Dynamics just released Spot 1.1, What advances can consumers expect to see with the new Spot release? (4:45)
- “The first software release that we had for Spot was designed primarily for this notion of remote sensing. So, you have a person who's driving the robot through an interface that is kind of like an Xbox controller. A really simple interface where you're driving the robot and you're able to see what the camera is seeing in real-time through a tablet or you can connect payload to Spot, but primarily using the sensing cameras on the robot to go through a space that you can't physically be in. That's really important for some of the customers that are looking to deploy automation in environments that are too dangerous for people to do inspection work. The example I like to point out is that we have been working with some electric utility companies that have assets and their substations that are too hazardous for people to look at while they're electrified. They have had these assets running for years, but they can't physically get close enough to them while they're running to see if they are arcing or if you don't understand what their heat signatures are while they are in operation. All that stuff. That's an environment where you'd want to send Spot. To go provide that remote ability to look at the environment. The 1.1 release was focused on adding new features that provide even more flexibility not just to the customers that want to do remote sensing but where repeated data collection starts becoming really important. We now have the capability for Spot to walk through a site and record the map that it has created as it goes through that site. It will remember the path that it has taken through an environment and then it can repeat that path autonomously without having to have a person joystick it through the environment again and again.”
How specifically do you see the Spot technology being used in the architecture and construction industry? (7:40)
- “A lot of the features that we are rolling out in 1.1 are geared specifically towards construction. Where the value is not necessarily about taking people out of a dangerous environment, but instead of being able to do repetitive tasks that right now or are hard to staff. Or, to staff in a way that is consistent enough. Where we started is this notion of remote sensing or data collection. One of the biggest challenges that construction sites have, as you probably know, is rework. You've got multiple trades running simultaneously on a job site. Sometimes even with the best-laid plans they don't always coordinate what they're supposed to be doing, and often there are clashes between what they're supposed to build and what the building information model says that they're supposed to build. There's been a push in the AC industry over the last several years to create highly accurate tools to compare the as-built environment versus the BIM of what the building is actually supposed to be like. In order to get the information of the as-built environment into the software to compare and do clash detection is really time intensive and is really poorly suited for people. Right now, often the methodology is to tell site managers, ‘Hey once a week, go around your site and take 360 images with this camera and then upload it into the Software System.’ Often that gets pushed back to the end of the week and if the person does it, it’s the last thing that they do right before they go on the weekend, and they're trying to rush through this so they can get home to their families. Often, the quality of data that they collect is not consistent enough to run computer vision-based analytics on the data that they collect because one day they take it on the right side of the room and one day they take is on the left side of the room. So, if you had a robot doing that type of data collection on a nightly basis and is able to collect the same data point again & again & again throughout the site, then you can start unlocking the value of these software tools that are designed specifically to help you track progress, prevent clashes, better-scheduled trades, track whether or not materials were delivered to the site. There is a lot of value in that type of repeated data collection. We are talking about 360 images here but the robot’s agnostic to the payload that you mount to it so you could add a 3D laser scanner or whatever sensing mechanism that site manager or GC feels is useful in getting the job done faster.”
We have talked a lot on the podcast about the future of architects and technology. Is Spot starting to create issues for the future of the modern-day construction worker? (12:10)
- “Anybody who thinks that people will be completely replaced by machines doesn't fully understand the limitations of machines. There are things that machines can do to help supplement work for people and these are the typical “three D’s” of automation which are dull, dirty, dangerous. It’s for things that are repetitive or really hard to do or potentially put people at risk. In terms of what we think of as the next step in the construction space right now we're trying to solve locomotion through a construction environment. Which is just the capability of walking through that space on a regular and repeated basis. That's really tough for most automation platforms because, one, you have an unstructured environment, mountains of stuff that pile up in the middle of the robots path or scissor lifts that get put into different locations. Having a robot that's not just able to walk over those but also to figure out a way to navigate around those obstacles as they change day to day is really tough. That's where we're starting. The next step is starting to think about how we can physically interact with a construction environment to start doing useful tasks. We've heard everything from material staging, so dragging stuff to the particular site where it is supposed to be for the next day's work. Doing tool retrieval which tool happened to have a tendency of walking all over the job site by themselves and so a robot is going back and retrieving those tools to make sure that they are restaged the following day. But then we can also start thinking about other tasks like doing layout or drywall installation and things like that which is certainly possible and within the realm of capability, but when we need to understand what are the requirements that are needed there. And, how far are we from actually being able to deliver a product there? Part of the solution might lay with Spot. If Spot the robot is able to go through that site on a daily basis and do some of this work, that is great but as we dig in closer into this industry, and all the industries were spot as being deployed, we want to understand what are the limitations, what are the high-value challenges that we can help address with our robot and maybe that requires a new morphology beyond four-legged robot like Spot.”
I love showing people the YouTube videos of robots like Spot or Atlas and people are really excited when they first see it, but then they immediately think it is a little creepy. What do you say to those people that think this technology is going too far? (24:10)
- “I was on this construction site in Japan about a year ago. In Japanese, the words scary and cute are very close or at least to my ear. There is Kowai which means scary and Kawaiii which is cute. You know when we deployed the robot at this construction site, site managers and people working on the site, came out to see the robot and interact with it. Everyone was saying it was scary. It was like a 70-30 split between scary and cute and that was day one. Day two, as more people started using the robot it was starting to shifting more towards ‘Oh really cute’ and less scary. Then by the end of the week, nobody was paying attention to the robot at all. It was operating as if it was just a part of the wallpaper. They understood what it is, and what it is doing for them and they just let the robot go do its thing. I think that speaks to something that is very fundamental which is that when something brand-new like this comes out and the intention or the benefit that is providing you is unclear then yeah it can be scary. What we've noticed is that people watch the videos, maybe they're freaked out, and then they see the robot in person and they get a chance to drive it and they are like ‘Oh I understand what this is. This is pretty cool.’ Then as the robot starts doing a repeated task that they understand and adds benefit to them then it starts becoming really quickly ‘Yeah, of course, we have a robot that's going around taking 360 photos of our job site. So what?’ It is just another thing. So, I think one of the biggest challenges in the space for us is just communicating the clear benefit that a robot like Spot can provide. I think that story becomes clearer to people. It will become less terrifying to something that's really exciting and aspirational. Then it will quickly become your why isn't it delivering my burritos faster?”