E89: Anibal Ollero, Full Professor at the University of Seville, Head of GRVC – Interview

January 23, 2017

Head of the group

This interview is with Anibal Ollero. Anibal is a full professor at the University of Seville where he is the head of the robotics, vision and control group. He also received his engineering and doctorate degrees from Seville.

Anibal’s biography is long and impressive. He is a prolific author, having co-authored more than 635 papers. And he has written 9 books including Intelligent Mobile Robot Navigation.

His current research is largely around advanced aerial robotics including coordination and manipulation. With drones becoming so popular, Anibal is probably one of the leading researchers in the world in this space, so we’re lucky to have him.

Here’s what else we talk about:

-What uses do you see for drones indoors?
-How did you get interested in drones?
-How do drones navigate indoors?
 
 
Transcript

Dave Kruse: Hey everyone. Welcome to another episode of Flyover Labs and today we get to talk to Anibal Ollero. Anibal is a Full Professor at the University of Seville, where he is the Head of the Robotics, Vision and Control group. He also receives Engineering and Doctorate degrees from Seville. So Anibal’s biography is long and pretty impressive. He is a prolific author, having co-authored more than 635 papers and he has written nine books, including Intelligent Mobile Robot Navigation. So his current research is large around advance aero robotics, including coordination and manipulation and with drones becoming so popular, Anibal is probably one of the leading researches in the world in this space. So we are lucky to have them today. So Anibal, thanks for coming on the show today.

Anibal Ollero: Thank you.

Dave Kruse: And before we get going on your current research, can you just give us a little overview on your background and what you did before getting into aerial robotics?

Anibal Ollero: Yeah. As you mentioned, I stated Electrical Engineering here at the University of Seville, but later I got some grants and I moved around. So I was in other Universities in Span, but also in France, in Toulouse and also in the USA. I was in Carnegie Mellon in the Robotic Institute and then I was working there in robotics. So that means that I learned from different environments and I also was a Full Professor in different Spanish Universities. I now have been collaborating for many – with companies. So I also learned from the companies about the project that they are interested and the application that they want to have. So that means that I am not only let’s say academic.

Dave Kruse: Got you, okay that makes sense. And what – how did you first start to get into robotics and what interested you in robotics originally?

Anibal Ollero: Yeah I was – when I was working in my PhD in the University of Seville, the department was related with automatic control and automation and so there is some robotics relation. But I still got later at Toulouse and later in Carnegie Mellon Robotic Institute where my main solicits of interest. I focused on information that also motivated my research in robotics, particularly my stay at Carnegie Mellon Robotic Institute was curtail for that.

Dave Kruse: Okay and what year where you at Carnegie – or what years were you at Carnegie Mellon?

Anibal Ollero: Sorry?

Dave Kruse: When were you at Carnegie Mellon?

Anibal Ollero: When?

Dave Kruse: Yeah.

Anibal Ollero: It was in ‘90 and ‘91, these two years. Later in the 90’s in the summer seasons I worked there for three or four summers. So that means that I maintained to contact on relations and for a number of years, yeah, in the 90s.

Dave Kruse: Got you okay, interesting and before we get into your current research, I’m curious, growing up were you a curious kid; like what where you interested in?

Anibal Ollero: Before? You mean the topics that I were interested? I always have been interested in filed robotics, so that means ideal robotics is relevant or the curtail, the most important topic for me, but before I were also interested – we were also interested in more of filed robotics topics like for example applications of lets say agriculture, forestry and so on. So that mean field robotic was also a topic in which I was working.

Dave Kruse: Okay. And can you give us – I mean you have a lot going on, in this space in robotics and aero robotics especially. Can you give us an overview of some of the research programs that you have going on currently?

Anibal Ollero: You mean in aero space robotics in general right.

Dave Kruse: Yes, yeah exactly.

Anibal Ollero: Yeah, well in Carnegie Mellon I was working in robots for a space exploration and so on. So that means that this was robotics, but let’s say now we can understand as let’s say wheels or legged locomotion that is not flying. But at the same time I started to explore this UAVs and ideal robots discussed. And also later I have been always working with let’s say aeronautic companies for example. I have been working with Airbus for many years dealing with application of robotics in the manufacturing processes, and in aircraft manufacturing. So we have a number of application also related to robotics in aircraft manufacturing, which is nothing that this – I don’t know if you know, but this is pretty manual. So it’s a not a lot of automation there. They need some kind of application of robotics in such a way that we can perform better, the aircraft manufacturing in the future.

Dave Kruse: Oh interesting! So you are even helping Airbus essentially automate some of their manufacturing by bringing in robots to do specific tasks.

Anibal Ollero: Yeah I was – I have been also working on the application of manufacturing robotics to aircrafts. I want to link with the idea of robotic, but this is more in the future, so we are also exploring how to apply in aircraft manufacturing aerial robotics. So would that means how to apply a small lets say in – because you know that they have few plans for aircraft manufacturing and they have a lot of problems related with for example logistics. So we would like to have these values in those varieties of robots, but this is the future.

Dave Kruse: All right, so you want to bring aerial robotics into the manufacturing processes to kind of help with it?

Anibal Ollero: Yes, this is what I’m saying.

Dave Kruse: Wow! Okay, so how can you describe a use case or how could aerial robotics help with the manufacturing processes, that’s really interesting?

Anibal Ollero: Well, this is quite different with many kind of UAV applications, but first up unique is something that can be used for exact, precise saturation of position and so on inside of the factories without having GPS and so on. But other aspects are related with safety for the people. So safety, but we need that different kind of UAVs to interact with the people indoor in such a way that they can still have some, let’s say possibility to fly without colliding or without in case of last colliding, then we can have some kind of energy obstruction for the UAV. In other words avoid, to avoid the hatches, to people and so on. And this is something that is – we need it for the future, yeah.

Dave Kruse: And with the – yeah I never thought about using UAVs in the manufacturing process. Would the UAVs carry parts or what would they, how would it help with the manufacturing process?

Anibal Ollero: Yeah, they can use, like its carries more parts. So that means that there is ready and a larger number of small parts that are needed in their manufacturing processes and then they can – these UAV’s and this special configuration can provide that too.

Dave Kruse: Interesting, okay. Well, that’s a good idea. And I’m curious what other projects are you working on that are little kind of future forward thinking, like the one you just described with the UAVs and manufacturing. What other projects are you currently working on?

Anibal Ollero: Yeah, this is also – well, first let me conclude that the first done project which is called aero culture and working on robotic challenges, which we propose to have this together with Airbus in their aero robotics challenges. So first we want to demonstrated navigation capabilities and positioning capabilities and so on. Later we will going to apply next year what there was in a use case okay, just to conclude this…

Dave Kruse: Oh no! That’s helpful.

Anibal Ollero: But now let me say that my main activity now would be and possibly an area of robotic manipulation and this is natural, because I have been working UAVs for many years, 20 years working on UAVs, but also working at the same time with robotic manipulators. Now they all have been – the last five years have been to have both together. So that means do we hire new manipulators that combine these two technologies, combine UAV technology and also combine aero manipulation, sorry robotic manipulation. So we are now the local aerial manipulation system with one or more arms to perform manipulation while flying for example, without need or for landing, just flying. This is one possibility that we are applying now and we are developing the, let’s say first aerial robots with one or more with multiple degrees of freedom robotic arms.

Dave Kruse: Interesting and I saw some videos that you have which are pretty amazing; which we’ll try to post when we post this podcast that you have up on robotic manipulation. So can you kind of describe like a used cases or one of the projects you are working on that – yeah?

Anibal Ollero: Yeah, in my case we didn’t love the very first area robotics with several digress of freedom arms or six or seven degrees of freedom arm for areal manipulation. With this manipulator we are able to avoid or to compensate the probation due to aero dynamic efforts and so on, otherwise you are not able to perform aerial manipulation, because having very special devices you can do something but in other cases you need more abilities to compensate this aero dynamic manipulation. So you can fly or manipulate at the same time, this was our case. But now we are working on IORMs, which is a solo project also a repeat on project in which we want to perform. First advance aerial manipulation with eventually more than one arm. So we are developing now aerial manipulator with a couple of arms in such a way that we can perform more advanced manipulation capabilities, but this is first. And the second is that we are also looking to invest in an application. So the idea is not only to develop this prototype, but also to apply in that field environment and one application that we want to perform is for inspection and maintenance. So that means that we need to be in contact, now just any data or photography with a sensor, but for example to have compact inspection in such a way that we can perform this inspection by maintaining the context with the pipes for example of the two phases or even to install a same source in the pipe in such a way that this insert can provide data later. Or even more we are deploying small robots verified in such a way that this small robot can operate out on emotionally for longer time that flying. So that means we want to transport and deploy the small robot in the pipes lets say 10 or 20 meters high on official site and then when they finish, we can again pick this small robots and transform back. Do you know what I mean?

Dave Kruse: Yeah, interesting. And so by having contact with the pipes, are they able to – will they have a sensor on them whether to test for I don’t know, would be rust or vibrations or vibrations or…

Anibal Ollero: Yeah.

Dave Kruse: Yeah, what would they be?

Anibal Ollero: Yeah, it’s very simply. Like for example ultra sonic analysis need to maintain the contact with the surf rate. For example, if there is any cracks on the lets say the, you want to analyze in the pipe how is the pipe in terms of the – let’s say how is the contamination and then how is the attack of the chemicals in the pipe. You want to see how it’s the weight, the state of the pipe in such a way that you want to measure the wall, how is the wall and then you can use this analysis by using ultra sonic currently which is another sensor which is electrical sensor to measure the walls of the pipes and then you are able to do that, but that means that you need the contact with the pipe. I mean wanted to have this contact we use this aerial manipulators.

Dave Kruse: Interesting. All right, so let’s dig into that used case little bit more because that sounds touch. So let’s say you are at a large refinery, I don’t know if that would be a potential, you know they have a lot of pipes at least. So you are at a large refinery, can you kind of walk us through from the beginning to the end how the UAV would take off and know the exact rout and then know where to test the pipes. Like how do you program that?

Anibal Ollero: Well, we have in the parameter experts in the end uses, and expert on how to inspect the pipes and then they provide the information for the plan. Then we have a plan for the aerial robot and then we know that it is interesting to perform some particular operation under the monitoring or under the supervision of the operators that we are at. But from the inspection plant, we prepare our plant for the roads and then this plant essentially to get the information that they want.

Dave Kruse: Got you. And how do you – what’s involved with that plan. So you probably have a flight plan, but how does the UAV know where to fly, like say in a refinery, because you know you can’t use GPS. How do you coordinate that?

Anibal Ollero: Yes, but not always. So that means first we need a very detailed map of the where the UAV is going to operate. So that means that in some cases this map is not accurate enough or there is some sections of the plant or the pipe that are not with an opportunity. Then we first, we need to fly and then this very accurate map. Only with this very accurate map we can plan this detailed inspection and yes, I have to say that in some of the areas we cannot have direct or good GPS signal. So that means that we need to apply a simultaneous localization of mapping technique, what we robotics people know like SLAM, Simultaneous Localization of Mapping and then you are able to navigate and at the same time build the map that we need. So that means that this is the first phase. What is built, the accurate map and this is needed for the intervention of the flying manipulator. And later you can always have some unexpected events and then you need to know how to react to these unexpected events.

Dave Kruse: Interesting! And how do you create that localized indoor map? I mean do you put sensors around the indoor space or is it more based on you know the UAV will understand, hey, I went to 10 meters this way. Now I need to – how is it – how do you build that localized map?

Anibal Ollero: Well, the first concept is just to local this site or to locate yourself with respect to their pipes. So that means that you need cameras and laser sensors in such a way that map what you have around and then when you have this and you have some known points, where it can be localized, at this point then you are transforming the relative local maps to more global maps and then you are able to perform these experiments locating the points in robot coordinate. But the first is to fly and to refine the maps that you may have from the beginning with more local information, in such a way that you are able to localize inside this, inside this plant.

Dave Kruse: Interesting! That’s a tough problem, that’s cool, okay. And okay, so after you create this localized map and the UAV has an idea of where its going, can you – how – and let’s say it’s going to go this pipe 1A which his 10 meters off the ground and that the UAV knows that where it needs to go. But how do you coordinate kind of the landing, and like with the vision system so that it can attach to the pipe?

Anibal Ollero: Yeah, the planning is really on hand. You may need more than one, because as you motioned the kind of flight of this system that we have now is not enough to perform the whole inspection. So that means that we have to plan and to perform some insurance or some inspection and then fly back and at the same time the other can take off flight to the other lets say to continue the inspection and so on. So this is not so easy and because also we cannot fly a lot of time and then that means that we need to combine the capacities of several UAVs.

Dave Kruse: Interesting ,wow okay. And how far long are you on with this research?

Anibal Ollero: We plan that before three years, we will be able to test the system in real plants. That means that we should be able to test, to start to test next year and we will finish this project in two more years in which we plan to have real experiments in industrial plants.

Dave Kruse: Well, that’s exciting. Well, all right, that’s pretty cool. So the vision could be to have many UAVs flying around plants doing safety checks and equipment chicks and pipe checks and that would be fascinating okay.

Anibal Ollero: Yeah.

Dave Kruse: So…

Anibal Ollero: I understand that oil and gas could be a more critical plant, because you also have to take into account the 8X certification and the explosive and so on, but this is not the only application. We are also inspecting for example bridges. So we are able to inspect infrastructure and we have another project to inspect bridges, so that means that you are able to fly under, in the undersize of the bridge and perform ultrasonic inspection on the bridge by the underside. So this is another project that we are also working.

Dave Kruse: Wow! Okay, how far along are you along with that project?

Anibal Ollero: This project is one GFO project. Again, we plan to have the very first experiment in 2017. So we will be inspecting some bridges in few months. The preliminary experiments will be very soon, so even we will have some preliminary experiments in the next weeks.

Dave Kruse: Interesting! With bridges, can you – I mean you are going underneath the bridges. Can you use GPS – I imagine GPS might be pretty hard to get once you get under a bridge. Are you using a different localization technology to map the bridge?

Anibal Ollero: You can use the GPS when you are approaching the bridge, but you cannot when you are below, because their accuracy, even if you have seen us from the satellite, you will have never, the accuracy enough to be out. So that means that you need to have other different positioning techniques under the bridge.

Dave Kruse: Got you, and that’s where the lasers and the computer vision can come into play and maybe a 3D model of the bridge even I suppose could help?

Anibal Ollero: That’s true, that’s true exactly.

Dave Kruse: Okay interesting. And so we are almost done with this podcast unfortunately, but I was curious, I know you worked with other companies and that you have transferred different technologies to other companies. Do you have any more examples of technology that you have transferred to companies in your past?

Anibal Ollero: Which one?

Dave Kruse: Oh I was curious if you have more examples of technology that you have transferred to companies that you have developed?

Anibal Ollero: Yes, we have been working in ground robots for example in order to inspect aircrafts. We also developed robots to be used by Airbus in their inspection of their aircraft in their manufacturing processes. This is something that also we did, but we have also in different fields, in which we have been applying for example and also we transferred, in the UAV world we transferred for example guidance navigation and control techniques in some companies and ground station. We also developed ground stations for some companies and we transferred this ground station to the companies in order to control the UAVs. All these are example of technology trends. As I mentioned before I have ordered technology transfers in the field of agriculture or the forestry.

Dave Kruse: Interesting! You’ve been busy over the years. Wow! You have quite the robotic experience and I wish I got many more questions for you, but maybe another time. But Anibal, I really appreciate you coming on our show and telling us more about your background and what are doing now. I mean you are really pushing the boundaries with UAVs, which is exciting and yeah, so you are doing all the hard work, you are doing all the advanced hard work, that’s for sure.

Anibal Ollero: Well, we have to combine advanced hardware and software. For example in this new field, which is in my opinion aerial manipulation, we can – we need to combine advanced hardware and software. We hope that in the future we will have also the integration of different other fields, like for example very light material. We need also more energy, because the energy limitation that we have now reset is very significant. So we need to let’s say, to combine our technologies with other technologies for the future for new applications, that’s there.

Dave Kruse: Interesting! Well, I’ll be excited to check out your progress in the future and see what you have produced. So I might check back in a year or two to see how things are going.

Anibal Ollero: Yeah.

Dave Kruse: And definitely. Okay well, Anibal I really appreciate your time and thoughts and good luck with the future of UAVs and thank you to everyone for listening to another episode of Flyover Labs. As always, I greatly appreciate it and Anibal, you can just stay on after we sign off, I’ll just do a couple of follow up questions with you. But yeah, thanks everyone, thanks Anibal for coming on.

Anibal Ollero: Thank you. If you need something more or you need something by email, just email me and I will try to answer if you have some doubts, okay.

Dave Kruse: All right, sounds good. Thanks everyone. Bye.

Anibal Ollero: Thank you. Bye.