Digging into key takeaways from our 2019 Robotics + AI Sessions event

Extra Crunch offers members the opportunity to tune into conference calls led and moderated by the TechCrunch writers you read every day. This week, TechCrunch’s Brian Heater and Lucas Matney shared their key takeaways from our Robotics + AI Sessions event at UC Berkeley last week.

The event was filled with panels, demos and intimate discussions with key robotics and deep learning founders, executives and technologists. Brian and Lucas discuss which companies excited them most, as well as which verticals have the most exciting growth prospects in the robotics world.

“This is the second [robotics event] in a row that was done at Berkeley where people really know the events; they respect it, they trust it and we’re able to get really, I would say far and away the top names in robotics. It was honestly a room full of all-stars.

I think our Disrupt events are definitely skewed towards investors and entrepreneurs that may be fresh off getting some seed or Series A cash so they can drop some money on a big-ticket item. But here it’s cool because there are so many students. robotics founders and a lot of wide-eyed people wandering from the student union grabbing a pass and coming in. So it’s a cool different level of energy that I think we’re used to.

And I’ll say that this is the key way in which we’ve been able to recruit some of the really big people like why we keep getting Boston Dynamics back to the event, who generally are very secretive.”

Brian and Lucas dive deeper into how several of the major robotics companies and technologies have evolved over time, and also dig into the key patterns and best practices seen in successful robotics startups.

For access to the full transcription and the call audio, and for the opportunity to participate in future conference calls, become a member of Extra Crunch. Learn more and try it for free. 

 

Danny Crichton: Good morning from Extra Crunch. This your host Executive Editor Danny Crichton. We held an enormously popular robotics conference in Berkeley, California, which we’ll be discussing today with our hardware editor Brian Heater and our Silicon Valley-based reporter Lucas Matney.

There are few spaces that are as interesting, as engaging and coming into its own as robotics. We obviously talk about a lot of future tech at TechCrunch, everything from Blockchain, to VR and AR, and 5G, but when it comes to things that are really just coming into reality (things that you can see, with products that are real, that are already on the streets in some cases) robotics is where it’s at.

So I’m going to introduce Brian who hosted our TechCrunch Sessions event yesterday, and Lucas who has covered the robotics space for a long time, in addition to AR and VR and other emerging technology topics. They’re going to talk about some highlights from yesterday’s conference, some lessons learned, and where the industry is headed in the future. So now I’m going to pass it off to them!

Brian Heater: Hello. Welcome to our call. This is exciting because we’re actually all in the same room.

Lucas Matney: Yeah, it’s a rare treat where we have the east coast and the west coast joined together by a shared love for robotics.

Brian: It’s just like after Tupac and Biggie died when all the rappers came together to work out their differences.

Lucas: It’s actually not that much like it, but Brian can keep rolling with that analogy.

Brian: So I wanted to kick this off by giving everybody a little bit of context and some insight the things that happened in the lead up to the event. There were a few things that happened behind the scenes that and when I kind of pulled back the curtain on a little bit and let you know, cause I think they offer both insights into the event itself and also a lot of context into the industry at large.

This is our third event. We did one at MIT two years ago. This is the second in a row that was done at Berkeley where people really know the events; they respect it, they trust it and we’re able to get really, I would say far and away the top names in robotics. It was honestly a room full of all-stars.

I think our Disrupt events are definitely skewed towards investors and entrepreneurs that may be fresh off getting some seed or Series A cash so they can drop some money on a big ticket item. But here it’s cool because there are just so many students and it’s a lot of robotics founders and a lot of wide-eyed people wandering from the student union grabbing a pass and coming in. So it’s a cool different level of energy that I think we’re used to.

And I’ll say that this is the key way in which we’ve been able to recruit some of the really big people like why we keep getting Boston Dynamics back to the event, who generally are very secretive. As you probably know, for the most part, the air, their only communication with the outside world are these videos that they put up every six or so months.

And I’ve developed a relationship with the company and Marc Raibert the CEO in particular, by essentially saying, “we have these events where it’s going to be a room full of very smart robotics students and either recent graduates or about to be graduates” and they are constantly looking for people to hire. And actually they are in the process of opening up their first west coast office in the Bay Area, which is the result of an acquisition.

Lucas can speak to this a little bit, but it’s amazing being in the green room and just watching all of these titans of this relatively small industry all talk to each other and to realize that a lot of these connections are being made at the event itself. And something that I said toward the end of the day when I was plugging our cocktail party that I unfortunately was unable to attend for work-related reasons, was “this is the one of, if not the smartest room that you’re going to be in. So absolutely use this opportunity. And a lot of these folks on stage are going to be out there, and it’s a perfect opportunity for you to ask questions or to hand off resumes and everybody’s just really happy to be there.”

Image via Getty Images / Michael Cohen / The New York Times

Lucas: It’s funny just seeing the caliber of the discussion because I feel so often, when it’s just a general interest conference, we barely get to dive into the meat of the topic because we really end up saying, oh, what are some of the challenges with AI some of the challenges with robotics, etc.

But here it’s an audience that’s like firmly dialed in, and that’s fun for the panelists and the speakers to just be able to make an assumption of where the audience is at and to be able to dive a little bit deeper on the topics. But back in the green room, it was funny because you see these folks like Marc Raibert very casually sitting and eating a sandwich while these also similar exact rock stars and other companies are just kicking out.

Brian: I realized something when we were dipping our toes in the water and trying to figure out whether we want to do our first robotics event. We flew a couple of us out to Boston and did a dinner. It was probably five or six months out and we were trying to figure out if there was enough interest in the industry, if there were kind enough people, if the infrastructure, if all these things were right and we had this little private dinner, it was probably 30 people tops, and it was very funny because everybody in the room already knew each other. And because it was Boston, basically everybody there had worked for iRobot.

I’ve got a ton of bullet points right here, which hopefully we’ll get through, so strap yourselves in. I’m going to kick the conversation off with these three key events that happened in the lead up to yesterday’s event that I think a lot of ways, obviously set the stage for the event itself, but also give a broader context.

So the first one is that about a week ago, Amazon made a key acquisition, picking up Canvas Technology, which was really nice for us because it actually ended up being this very roundabout way to get Amazon Robotics on stage. So Canvas is a logistics and robotics company out of Boulder, Colorado. They essentially build these self-driving cars, like a version of an autonomous vehicle in the warehouse. So they drive back and forth, and you can put things on them, and they require minimal human contact.

I saw their device in person about two years ago when I was visiting Playground Ventures in Palo Alto. They had an open house and they had a lot of their products around. One of them was this Canvas robot that was driving home, and delivering drinks to people. It was very impressive and it was doing an amazing job of waiting on people. Doubly impressive because it was company that up to that point I had never heard of. And a week ago when Amazon, when broke the news of this acquisition, I don’t think anybody had really heard of them. Lucas, are you familiar with them?

They’re an exciting company, and we decided they’d be good for this “How to build robotic startup” panel that we kicked things off with and which Kate Clark moderated. We actually ended up also having one of their key competitors, Fetch, and a much more prominent company. It’s extremely impressive technology. And it makes a lot of sense as an acquisition for Amazon.

Basically, Amazon has been building a fulfillment center robotics ecosystem since about 2012, which turned into Amazon robotics and they’re basically doing it piece by piece. But, again, a few months back, myself and Ben are our video producer, spent at day fulfillment centers and got to see these really impressive robots that they have. They’ve got about a hundred thousand robots deployed across 26 fulfillment centers. So this is seems like a no brainer. It’s a perfect fit for them.

Getting back to Boston Dynamics, they also made a big acquisition, which was actually their first major acquisition. They bought Bay Area-based Kinema Systems, which is a 3D-imaging company. And it’s a thing that Boston Dynamics can do when they’re owned by somebody like SoftBank with extremely deep pockets who I suspect is probably playing a pretty big role in pushing them into actually commercializing these products. The acquisition is very interesting from what it actually means a lot for the future of Boston Dynamics.

So like our stage last year, Marc Raibert was there again. But, as they announced the Kinema thing about a week or two ago, they also announced that they’re going to be commercializing Handle which is this crazy kind of looks like an ostrich, maybe one of the most unique look, which is saying quite a bit. It’s got two wheels, it does it really like impressive balancing thing, it’s got this really long neck. They showed it off last year, and it wasn’t quite clear what the deal was.

They’ve completely retrofitted this where they’ve completely revamped it with all the Kinema technology on top. So it’s got this 3D-imaging system and at the end of this long neck or arm, it’s got a large round object with a bunch of suction cups on it, and it’s just doing this very impressive, picking and placing of the boxes. It was funny watching Mark talk about that because you can just tell he’s the guy who would just build a company forever and never want to productize it.

Photo by John Phillips/Getty Images for TechCrunch

I don’t even think it’s a question that they felt like the most impressive robots. Boston Dynamics was building these robots for the military. They didn’t work out because they were too loud cause they were hydraulic based.

But they’re commercializing the robot and Kinema is going to be a very big part of that. Kinema was building robotic arms, which a lot of people are building, but they have these really advanced vision systems that are just really good at identifying objects and grabbing them. And Boston dynamics had built some stuff internally, but they’re not a vision systems company. So they’re doing a lot of off the shelf stuff and they bought this company that specialized in that specifically for this product. He alluded to the fact that they’re going to be using it on other products going forward without mentioning specifics.

But, it speaks to another really important point here, which is the kind of the pragmatism that’s required to run a successful robotics company. Boston Dynamics has far and away been an outlier. They’ve been going strong because he had the grants for a while, they got acquired by Google who didn’t know what the hell to do with them. It didn’t have anyone to push them in a good direction. But it also feels like; I guess if you just had massive cash reserves, why would you not want to own this company?

Brian: All right, well let’s get to some questions. Number one: “The AI monitoring systems described in the context of autonomous driving sound exactly like surveillance systems. Can you comment on how that information will be kept secure and used appropriately?”

Yeah. not to get too dystopian about it, but I think the fact of the matter is that everything already has a camera and things are going to have cameras on them. So I don’t know what specifically, outside of drones and things that are actually made specifically for surveillance, I don’t know that things like vision systems on autonomous cars are going to be more of a privacy concern then any of the other surveillance that is happening with autonomous cars. It’s not in the best interest of these companies to have this stuff go to the cloud. It’s like most is going to happen on the edge anyway if it is to be trusted like with vision systems that we’re seeing. It’s a lot of 3D mapping models.

Obviously, the reason that these cars are getting better is because some of this data when these cars have problems is getting fed back to the cloud. But I mean, for the most part, the technologies that they’re using are not necessarily giving photo realistic representations of the outside world. It’s their point clouds. However, if you look at how much our lives have been improved by Russian Dash cams, Internet content, maybe it won’t be a bad thing and again, at the end of the day, all of the surveillance that exists, I don’t know that I would be more concerned by drones flying everywhere.

Lucas: True. I mean the thing is it’s crazy how these companies can’t even grapple with how to hold on to our two-factor ID phone records. So the only thing that makes me feel better about it is that it’s just a massive amount of data. So how is it really in their best interest to save several terabytes of someone’s driving data all the time? That’s just not feasible for them. And that’s, I guess all that being said, that was my initial thing about like, well why would Amazon want to keep all of these Alexa records because that’s just so much data.

Privacy is going be an ongoing deal. I think maybe these companies feel like they don’t really have to chat too much about it publicly right now because A) they’re figuring it out and B) most of the pilot programs that they’re engaging with are very closed in terms of having their own employee driving the cars. We’re still gathering up city data, but it’s a different beast right now.

Brian: I think the thing that became very clear to me in the past several weeks was, we’re talking to Uber and all these people working in autonomous cars, and the expectations in terms of the speed at which these things would become realistic options, it seems like everything has been pushed back. I think all of expectations have been pushed back. And a big part of that is it’s not just privacy concerns but safety concerns. I mean it’s going to take a lot, a lot is going to happen to get the general public to really trust having these things on the road.

Lucas: Honestly we could talk for hours and hours about this because it is a major concern and the amount of data that these cars can grab is just crazy and it impacts everything from the mapping systems to having 3D point clouds of the worlds that feed into AR capabilities.

Brian: So the next question relates back to what we were just talking about with Boston Dynamics: “This is my personal opinion, but Boston Dynamics seems like a show pony without a business model. Relatedly, haven’t we learned from Jibo that a slick video is not the same as a working product?” Yes. Awesome.

Image via 6 River Systems

Brian: It’s a very interesting time for them because something that Mark talked about on stage was that there are basically now two sides of the company. I suspect that they inform one another, but there is the far off research side of the company that’s making new orders to really forward the field and make really impressive robots. They’re impressive videos, but they’re also really impressive robots.

They’re forwarding the field, and I don’t think you can make an argument that they aren’t, but there’s going to be a lot of growing pains and it’s going to be interesting to see how they make the transition into commercial products. Whether they can keep the pricing down, when people want to avoid surveillance. I think as we discussed Boston dynamics really hasn’t had any internal force for a while. Pre-SoftBank acquisition, pushing it to heavily productize and move that direction.

Lucas: Yeah. In these warehouses, big companies are willing to spend a pretty penny on these products, but they really have to get that price point right. And when you think of, well actually I don’t even know because like obviously what Canvas is working on just kind of seems to more seamlessly fit in with the existing warehouse scenario. Whereas this is really going on with the expectation that there’s maybe a human in the loop looking down at these robots. But really the key difference is the fact that Canvas and Fetch are very minimalist robots, they’re real. They’re built specifically for a purpose. They’re modular. They can do more things now. I guess that’s going to be the hardest question. What I did think was interesting in your chat with Mark was you talked about SpotMini as a platform and called it a modular platform.

And the modular platform examples that were given in the videos were very much like I could see the same thing coming from like an RC car that could have a 3D thing. But I guess in terms of what is it a platform too? Like DJI does the same thing with drones and then they decided this is the engine-based company that is able to create things at scale.

Brian: I wanted to touch on something really quick and then we’ll get to our third question from a reader named Max. This is the third of three things that happened to me. This one is the biggest bummers as well. So a couple of weeks ago, one of the three companies that we were planning on having our drone panel went out of business. The company called CyPhy Works pivoted from hardware into AI data collection in the drug field and then seemingly overnight went quiet and closed up shop. I was scrambling, to get a comment from the comms team, and the comment was quote “Aria Insights, which is the new name of the company has ceased operations effective March 21st, 2019 period,” which is, it’s a very flattering comment but its basically “we don’t have any context to give you.”

Lucas: Or even like “we don’t actually know what’s going on right now”. And so this is important for a couple reasons. Number one from a personal standpoint, it left us scrambling, but we are well aware of the fact that hardware is hard and that robots are very hard and there are a lot of moving parts and doing events like this. But the other key point here is that the company was cofounded by Helen Greiner who was one of the three cofounders of iRobot. The second person being Colin Angle, iRobot’s current CEO who was with us on stage yesterday. And the third is Rodney Brooks of ReThink Robotics, which is a cooperative-assistance robotics company.

A lot of people were using this research company and Aria basically went out of business overnight. They tried to find a buyer, apparently something fell through last minute and they completely shuttered the company. So they were both very well-respected and very promising in the field and other one and was able to make it in spite of all of these things they had going for them. And I think that this gets to the Jiba thing from the standpoint of that speaks to the amount of pragmatism that’s required to run this successful robotics company.

And so you pointed out the fact that it took them 12 years to get to the robot and the path to making it so if you visit their offices in Bedford, Massachusetts, they have a little hall of fame in the front and it’s all of the things that they attempted to create before they got to the robot. There’s a Mars Rover they were doing, there’s all these baby dolls. They even have the Roomba. He has this pretty good joke that he told us where he says that he wasn’t successful roboticist and until he became a backing salesman, it’s a good line.

Image via Getty images / Matthew J. Lee / The Boston Globe

Brian: One of the things I thought was pretty interesting about what he said was he was talking about how a couple of decades ago, looking at a vision-based navigation and some of these very wide cross-industry problems. Like those were robotics problems exclusively and how just over the past decade they just invaded other spaces. And so all these massive companies have had R&D resources directed towards us, and now a lot of this stuff is easily available and open source to the point where you can just start a robotics startup in a lot simpler ways than you could have a couple decades ago, but at the same time you still have to like being in a spot where you can be flexible and in a way that iRobot seems to embody more than any other startup.

Several years ago now when I was with Engadget, we went to the robotic hand lab at Harvard. They’re doing some brain interesting research out there, and what they were trying to do was to create really inexpensive hands and they had worked out a technology where they would take a barometer which had been manufactured on the chip for phones. All phones have barometers on them. And one of the reasons is it’s a way to triangulate position inside of places cause you can measure changes in the atmosphere. So all of these other technologies, screens and batteries and everything else, all of these companies with massive R&D like Apple, Samsung were, there’s this big arms race to enter and the price came down on everything, and these components became super cheap.

They basically did some of these with silicone molds and made fingers and it was a way of detecting pressure. So the hand would grip onto something, and it would use the reading from the barometer to tell the change in pressure. So it was like a very way, pretty cheap way to make a robot hand feel. And I think what Lucas is getting at and, what Colin was getting at with Amazon Alexa is, I mean like Amazon’s push to make this smart assistant and Google’s got all these like knowledge maps that they’re able to access artificial intelligence, all these other things and they’ve become a lot more cheap and accessible because of these consumer products.

Lucas: You can think about that and then think about how much longer it might have taken, some of these top research institutions to come up with the early inklings of unsupervised learning techniques had they not been forced to moderate the content on their platforms. These are just, their custom problems that no one would have expected would have been the thing that would push this forward. But it’s just how it happened.

Brian: Now, the question from our reader Max: There’s a lot of activity around robots that work within warehouses – have you seen any interesting solutions for transport between warehouses, especially in large distribution centers (Walmart, FedEx, Amazon, etc.)?

Lucas: Yeah it kind of falls into general autonomous situations. I’ve been going to the Y Combinator demo days it’s been interesting the past few classes it like seems there’s always been autonomous shipping startups. Cargo is a startup, which has just been interesting. Like how many competitors is really YC betting up here. But that’s, that’s obviously a little bit of a different situation.

But basically, at the heart of this question, when you look at it like self-driving trucks between warehouses like that is what’s going to be the situation. But then you look at loading the pallets and getting the pallets on the trucks, that’s a solution like what Boston Dynamics showed off.

Brian: All right. I’ve got one more anecdote from the event and then we’ll wrap this thing up. Okay. So related to the I robot thing. So we had a company called Breeze Automation basically come out of stealth onstage yesterday. They’ve already got contracts and actually based around the Bay Area. They already have contracts with the US Navy and NASA and are doing some very cool things with extra soft robots made of fabric and air or water.

So the founder of the company that we had on stage. I’ve known him for a few years since he worked in a maker space out in Massachusetts, but we had met a tech event in 2015 or 2016. He was on stage with the committee called themegabits, which is knowingly chuckle.

But it they had this very impressive looking robot that was parked outside pier 70. And it was like obviously a big giant show that they were putting on. They built this big robot and won one competition with the Japanese team. And then after that, the company just sort of just disappeared off the radar.

And I’ve been like spotty communication with Gu, the founder, but I met up with him a couple of months ago for the event and recruited him for this to talk about what he’s working on. And it’s impressive. They made a lot of progress just from the standpoint, and they’re actually like making something that the Navy and NASA are interested in, working in these like very extreme areas and, they’re making a lot of progress. So again, I think that speaks to the robot thing and the Jibo question, all of these other questions. Just taking a more pragmatic approach and actually trying to solve an actual real-world problem with the device.

Lucas: I didn’t get a chance to see it. So what was the form factor of what they’re billing?

Brian:. It was an arm made of fabric actuators. The whole thing was fabric, and it inflates if you’re doing it in space.

Lucas: Carbon fiber fabric or what?

Brian: I don’t know of specific fabric, but it was like, I mean I don’t think it was carbon fiber. If you’re in space, you can feel cold air and if you’re underwater, full of water and it works on the principle. Same principles apply to these other software robots in that the air and the water are like doing most of the powering here, right? It’s like based on the pressure you can open and close the hands and all these other things that you should check out.

Image via Getty Images / Jeff Swensen

They showed us three videos and one of them was a telepresence where the guy was sort of controlling it remotely, and you are using it to pick up some of stuff, but it was really impressive technology. But again, I mean I think I touched on this last time, the real break there is that we’re going to start to see in robotics are the way less sexy ones, those behind the scene ones are the things that are happening at like Amazon fulfillment centers or all these other things coming a long time before all of the consumer robots.

But hopefully, probably, in the same way, NASA’s situation can be the the iPhone creative barometer for a robot hand. Probably a lot of these technologies that we’re seeing right now have behind the scenes money where companies like Amazon are pumping millions of dollars to create something that can get you to your box the next day will eventually filter down.

Lucas: And it just seems like in the past couple of years, like if you like look at the videos online of these warehouse robots, not only interacting with the merchandise, but there’s a network of them interacting with each other and it’s just mind-boggling. And when you think of robotics, you don’t think of it the same way anymore just because it just seems so futuristic and, I don’t even feel like when I was thinking of this event, I had the consumer applications really in mind because this just seems like it should be so little, since this is revolutionary and transformative for warehouses right now.

Brian: Definitely. So what was your favorite thing from yesterday?

Lucas: Obviously the energy in the space was generally pretty happening. But I think, particularly for your iRobot and Boston Dynamics panels. Even where you have veterans of the space who are just leaders of these companies like Marc Raibert rolling around in his Hawaiian shirt and the iRobot guy just being so perfectly candid.

It’s really the perfect event. So often when you’re talking to a CEO, you can’t really tell what their genuine opinion is and what’s really coming through. And I think that the people that came to this event really got to see some geeks that were really down to just share their insights. And I thought that was pretty tight.

Brian: I think the difference between this and say, we did the AR event last year and it was like there is this – obviously, AR is in on the ascendance but sometimes you’d have a panel where you’d have to ask the question of like is this trend still on? Like is this still happening, whereas Robotics just feels like the right time for it. This feels like everything is accelerating.

Lucas: I think one of the interesting and consumer level thing s was how iRobot showed off the mower and stuff. Which was very cool. And the way that Colin was talking about the solution to why it took so long to get the mower out there. It was pretty equally interesting.

Brian: Right. And it’s just you look at iRobot vacuum, and as matter of fact, they are like well you’ve got your big floor in the backyard and like going through some of the process. I thought that was pretty interesting in terms of their pitfalls and challenges.

So iRobot is based in near Boston and I was there this time last year to talk to Collins cause we had him on stage at Disrupt, and they have this little like to have that front where they used to own where they would do testing. And it was the same company, and they came three different things. So he would do all the testing out front of these military robots and another one that did like bomb scans and things like that. And they told me that when I was out there and they were like testing the lawn-mowing robot in that area, and just nobody was any the wiser, but that in order to test it in places like Boston, they had to find greenhouses. So they would just have it there, like these great houses with this secret robot they were testing that was just cruising around, and people would walk past. And he said that they – he basically alluded to the fact that they had another significant product and that they were thinking about sharing. I joked about an iRobot snow blower because it could really cool to have like a little black blower thing.

Well, all right, thanks, everybody.

Lucas: Thanks, everybody for joining us. We appreciate you being Extra Crunch subscribers and tune in cause I’m sure we’ll have a ton more of these. Thanks so much.