Gadgetory

hi I'm Paul Miller with the verge I'm here with Mark chiappetta the vice president of advanced development at iRobot and you built ava is that correct that is correct with your own bare hands alright not exactly but but this is a hand-built prototype of a mobile robotics platform after you what is ava so ava is the world's first practical robot app development platform she takes the best innovations from the mobile computing space from the electronic gaming space and of course the mobile robots industry and brings them together to bring robot app development to the ios and android developer Oh communities and so there's a lot of technology on here but what you guys want to emphasize is the apps that can be enabled by this yes so the application development is definitely one of the key missions of this robe on you know exploring new market opportunities for us internally and developing partnerships to solve some practical you know robotics problems but we're also using this platform to continue to move the state of the art in robot navigation through real-world environments right so you guys are working you guys have a partnership with Google right for app development it's correct what how is that going what's what's happening there there's a couple robots out there that are having applications you know developed so Google has a couple Ava's yeah that's correct okay and what what are they working on there oh you know I really I'm not sure android I vid yes Android and I I think they're sort of exploring the space at the moment okay and is that is that the what are you working with other developers that are trying to figure out where they can fit in so you know right now we just struck a partnership with In Touch health to bring a derivative of this platform into the healthcare space allow doctors to do remote patient monitoring and rounding is that research or is that a product that you guys are working to deliver no that's a geared towards product development so let's just talk about Ava you mentioned it's got a this is basically a kinect sensor what is the pure direct it's a prime sensor crime sins there are actually two of them and the prime sensors are depth imaging cameras where as well as RG where it was the other the second one is uh is under the shoulder and it pans back and forth and it looks at the volume in front of the road so it's like an IR sensor but it's just panicked so it's another connect or connect type of sensor oh there's like that hands in front to develop situational awareness and to avoid obstacles that are in front that way you guys were developing this like about two years ago started yeah we started around March of 2010 was this even available to you at that point the platform itself the these sensors um yes it was but in prototype form because the Kinect hadn't been out yet right right so you guys had that pretty early and it seems like is that I mean we for the hacker community that's been such an enabling technology but it's so cool I mean even I robot is using this sort of technology yeah you know I robata is uh we have a number of core competencies from one being over all that navigation other one being building you know practical products right we also develop our own sensors in house but we that's a shared competency I mean we are always looking outside to see what the latest in you know high fidelity sensing what is around so for example this has a long-range scanning laser rangefinder right and it's based that it's used for localization and mapping and it's got some sonar in there yes it has upward looking ultrasonic sensors to prevent the robot from you know decapitating itself on a table but we got the boot that's a smart that's a good one and we got to play around with it outside I mean it's you know with the brake collision avoidance and you guys have a nap here you can just you know point to where you want it to go and it goes there that's correct there's a number of ways to to get your input into the system one is there's a map you can click on a map and the robot will autonomously plan a path to whatever that point is you can string points together to make like patrol paths you can also click in a video image and it'll go to where you click in the image you can control the individual degrees of freedom right by touching on the image if i wanted to make it say no way or you know or it she also responds to you know direct right touch as well so like i can like and look at me what was it you know if you uh if you do that she'll turn and then tapper goes the direction she'll turn towards you that's the one I was doing this so well before I profit and obviously you could if you bump into it with those pads down there she'll stop blush correct and but this is this is how users are going to control this do you want app developers to have access to this right I'd use that to navigate yeah absolutely we've created an API and we've exposed it to the face device which in this case happens to be an iPad but can just as easily be an Android tablet and you know the sky's the limit right this is a very robust safe and you saw it move it's very Swift you know navigation platform that is equally at home in a home environment in a business environment or even in a healthcare institution where the operating around people and accomplishing its mission getting from point A to point B is you know it's state of the art doesn't sound pathfinding absolutely it'll plant its own path and using the it's a knowledge of the world in a map that it's created by itself so it does create that like now right now you guys kind of have to like skin like set it up a little bit but will it be able to dynamically generate its own map for things yeah absolutely what you working on autonomous exploration as an example okay in their research space so that you know you could pull a row but out of a box right and you know fire up a nap and say map and it'll drive around your home and boring you're in your office space and will create the map on its own okay um and you know this is an engineering development tool right here so it's not very pretty right obviously the maps can be post-processed to give the user you know a really nice-looking user interface and you guys are doing stuff this is similar to your military robots it can can follow you is that correct that's correct yeah this robot on does person detection and person following my face recognition um not facial recognition but it looked at the upper torso okay in the case of this robot and identifies that it's human being and then can either follow you or go away from you or go you know whatever its program so what's what's next for for ava is this early development where would you where would you say you are right now I'd say it's market exploration early partnership development and you know continuing technology development for us right and what's next you know greater exposure to the developer communities would be real is of importance to us is that a messaging thing for you guys or is there stuff that you need to do to be ready for a lot more developers to get involved I think we're putting we're building the groundwork to engage with greater numbers of developers so there's still a little bit of work to be done there and we have limited numbers of these prototype platforms now soon they'll be derivatives right in the market place under the intouch agreement right what can you explain that a little bit sure we saw we signed an agreement with In Touch health or low here that one that one in the year yeah to develop on and that will be in the market when we're not saying exactly when it'll be in the market okay but it's we're in joint development right now okay great and are is what comes in the market going to look like this I think you'll see the family resemblance okay but it won't look exactly like this we'll have limbs ah oh I would say that one will not have limbs but the health want to help 11 on Italy but are you guys looking into that researching that yeah we have a lot of manipulation development that's happening in house in research and you know right now the problem of going through cluttered environment safely is solved right it's solved in a way that you know obviously good enough for highly cluttered you know even small environments where so you can imagine next steps would be okay now i can get from point A to point B if somebody puts a meal tray on me and tells me to bring it into the family room it'll do it right but it can you know manipulate the world yet itself so that's then the goal is for it to be able to do that that's just one possible application that could be developed with the right hardware what's right hard right hey that's I mean that seems like the that's been a for a consumer robot outside of these single purpose ones you've done with like rumba and the gutter one luge yeah you know the idea that a robot could go to the fridge get out a beverage of your choice and bring it back to I mean that's kind of like that the idea in milk a soft drink exactly uh that's kind of been the I don't know for some reason that stuck in my mind as we're consumer robotics need to go for someone to spend what this would cost to get it in their home a single purpose robots are like with the Roomba you know obviously have great utility we have a vision of an automated home where an Ava type robot is the focal point and it's there to interact with humans so that the humans don't have to interact with the other application-specific robot right so this robot is here to command the other robots to to serve you and it's also here to directly you know serve you in that vision so absolutely going in you know fetching and carrying things for you you know allowing you to be somewhere where you're not physically delivering content to you through you know web connection and her tablet computer so those are all potential applications and are some of the barriers there the actual technical complications or are some of them the cost implications of bringing something like that to consumers so you know that's a great question so we're the leader of you know the practical robotic space and you know you won't see I Robot more than likely release a product that doesn't have a very high you know benefit to cost ratio right so the Roomba is you know for what it can do you know it's price reasonably low but um it's it's really both right there's still technology development that needs to happen and also there needs to be market pull right there needs to be a practical problem that needs to get solved at a price point that we think we can solve it in for us to bring a product derivative of this to market but you know people didn't think they needed a room bow before you had the Roomba right um you know that's a classic problem in the field of mobile robotics right it's in the early days you know over the last like 20 years folks would call mobile robotics you know solutions running around looking for problems and the Roomba was really the first problem that customers had they knew they had they just had a different way of you know solving the problem a more conventional way with uprights and canisters and whenever we talked to people they would say make me a robot that vacuums your floor's right that was like the first killer app you know getting a beer out of the fridge that's pretty much a man's you know first right that's an hour of vacuuming versus a minute of getting out the cow exactly I see that now you back to what you were saying about interface is there an aspect of this that the robot is is user interface even more so or in some ways different than just the practical application so ava is capable of interfacing with people through speech through gesture through touch and as advancements continue in the mobile computing space things like facial recognition as you mentioned before or you know basically other applications in that area right we should hopefully leverage those to deliver those advanced capabilities do you think of that as natural user interface in the sense that like Microsoft things with the Kinect or is the fact that a robot is a physical present is that a more natural user interface um I think it's both you know you know to take connect as an example playing tennis with a Kinect is it's nice it's interesting but there are some times or you want to actually physically interact with something right you're not holding anything in your hand you know steering wheel you're doing this right right so the tactile inputs are very important to us you know robots are going to be in the environments with people people are going to touch them you know it's it's a common way you like them up to you and tap you on the shoulder you're going to turn around look at me one of the ways that humans are used to interacting with other organic beings for robots we want it to be the same yeah and our developers getting excited about that you have like UI designers that see it that way I think so yeah we uh there's a number of people both internal and external that quickly see the possibilities you know the possibilities are pretty much endless you know and whether I robot or not does it when when should I expect something like this when when am I gonna get the beer out of the fridge let's submit I want a beer out of the fridge is that five years is that ten years yeah that's close you know like i said you know whether if you're talking about in a product it's farther out we're talking about in research it's happening now yeah right the question is you know is it something that you'll want to spend your hard-earned money on you know is it available at the right price point that's a little farther away but you think most of the technical problems at this point are solved the visual object recognition grasping all you know those problems are still being solved you know many technologies feel like they're solved where they're done in research but then when they're they end up in practical environments or which wrecked but you know randomly a very bro you find out that those problems aren't quite solvent it takes a number of iterations of being in actual in out in the world were to have them quote unquote salt which seems to be an excitement citing thing about Ava is that you've got it out in unstructured environments already absolutely hurt her navigation ability through cluttered environments is unparalleled we don't know there are other robots that can navigate similarly but a lot slower a lot less dynamic you know she's also as I mentioned before holonomic meaning that you know she's capable of moving in any direction while facing any orientation like strafing she could exactly like a first-person shooter right she could strafe she can orbit she can sidestep at two meters a second when she's going down the hallway if somebody jumps in front of her so she's very very sort of deft is that why we're going to see most of these things on wheels for the time being yeah legs if you're referring to legs yeah you know legs are obviously there's a ton of research happening there's there's toys and products with Bill the flags but you know legs are legged robots are still being held back by how much power they can put on board and the complexity that is there and complexity equals you know lower reliability and greater cost is just not it's just not cost-effective and it's not necessary I think you'll when you see robots you know climbing stairs which were developed for people to change elevation right then you know you may see legged robots so it's something you guys are aware of right I'll keep your keeping tabs on you know absolutely a leg locomotion I think where you'll see you know practical implementations of legs are more like in you know prosthetic applications yeah sweet well this is cool I'm excited I hope to see it watering around the show floor soon thank you will probably but and I hope to have it in my home as well yeah you'll have to talk to Lauren oh I will I will definitely and then once you on meant a robot arm I think I think you have something here alright thank you so much you're welcome next fall