 Hi everybody. So, before I talk, I'm going to give a first introduction about myself. My name is Sunwe. And I'm an advanced development scientist in ASTAP. I'm an advanced manufacturing and technology centre. So, I just joined the RFC like three months before. I worked in the MDU, a robotic research centre for two years. And I'd very like to join the ASTAP industrial research programme, lead by Professor Marcelo. So, in the past two or three years, I've been continuously working in ROSS and ROSS industrial. So, also, you must be a part of the company. So, we really see a great opportunity in this domain. So, I think it's a great future for also the developers, startup builders, makers, like them, yes, research. So, today I'd like to take this opportunity to share my knowledge and some information in this domain. So, the presentation type of the day is how open source robotics is accelerating our research and innovation. So, before talking about open source robotics, I'd like to take the opinions in the computer sphere. And I'm joining the smartphone industry as an example to justify how open source is important. It's a fundamental change in our life in the last ten or ten years. So, this figure actually shows some data that nowadays, also, there are only 5.5CM shares on the desktop computer for Linux, but almost half of the server run Linux. And for smartphone, as you can see in this figure in the last eight to ten years, 88CM share of smartphone already running Android. So, by the way, Linux and Android are open source projects. So, in total, the Android system is like 37% of max share in the whole operation system. So, this is quite a main story to show how the open source is quite important for product and for innovation. So, today we're talking about open source robotics, the two sides, hardware and open source software. So, the open source hardware has already played a key role in the past years to help our development companies to be able to develop faster, cheaper and better. A good example are affordable, like 3D printers, lights and carpets. In the open electronic kits, like Android, the Raspberry Pi, we use 12 chips that are easy to use. And it's open source, so different manufacturers can have different versions. All these design PCB boards are already online. You can even design your own hard unit boards, have your own development environment. So, for the open source software part is obviously already widely used in industrial and non-industrial. So, it's quite a success for package in the computer vision area. Gazebo is kind of a dominate simulator in photo robotics now. So, it's also a part of the OSI vision. So, ROS actually is the dominant open source software for programming robot. The other one's job for using ROS is you don't need to reinvent the views. This is the concept. So, your works will be on the success of other things work. So, it's very quick and easy to be at your own application. So, just a little bit about ROS, probably most of you have heard of it. And James just did a very good explanation about this fantastic project. But I'd like to also expand a little bit about ROS. So, ROS is a kind of open source operating system for robots. It's originally from Stanford. So, a project. And then the village garage is a kind of incubation that they stay committed. So, the tremendous work helping to build a ROS kind of widely used software, open source software in this community. And the core concept of the ROS is it helps you to break a complex software to small pieces. Take example, if you want to be a robot, service robot you have different kind of modules, push-up planning, exception, task generation. All these kind of things you cannot do by yourself. So, you probably need to take out things or modify them in your own robot. So, ROS is the driver of the development to reuse other people's work. Of course, the license. Always there on the top. How is that in the end of my presentation? So, ROS provides a kind of a framework, source and interface for distributing development. So, for a lot of project people we can have her to contribute in the different corner of the world. So, in general, for our community developers we put our source code on GitHub. So, we collaborate together and we have this project. So, this is a fundamentally new way how it would be a project and make this stable and reliable. So, in one sentence, the ROS is a planning plus tools plus different capabilities. And what's important is the ecosystem behind of it. They are like thousands of developers. We have a week ahead of us. So, you can ask questions and they will be expert to answer your questions. So, just some data and the figures show that how great ROS is in the past nine years. It is going pretty fast. As you can see, you need to figure some data in the past ten years. The top line of the program is like 14 million. This is equal to like 1,400 authors programming at the same time in the past ten years. So, they are like four millions. You need that of course over there. And to get the idea of how big this project is, if we say this is a project running by a company or a research group, this is like 1,236 percent year of development. And this is like 137 employees work four times in the past nine years. So, we have never seen any robots project or software project that you put so many human hours and time in it. So, this is great. Since ROS is originally used for service robot, because we understand for the project that has a PR tool, personal service robot. But the ROS industry will actually extend the mass capability of ROS to new manufacturing applications. So, it's more reliable, stable. And we have industrial base to come in and contribute it. So, the ROS industry factors can do, for example, collision avoidance project generation, automated adaptive tool pass generation. This is the software that the robot manufacturers like KUKAI have never brought before. So, they are the industry of the good at doing fixed pass repeating task. But they are not good for collision avoidance trajectory, even industrial calibration not make it. So, ROS industry brings this new capability to the traditional industrial robot. And this is a requirement raised from industrial 4.0, which means robots should be flexible, should be able to deal with low-value, highly mixed tasks. So, I think the ROS industry will play a great role in the future. There are other process like finishing a building, this traditional one, patient robot process, still ROS can play a great role. And we already have several success famous. So, there are new areas like aerospace, logistics, even house, care, domain, it can be used for more development. The ROS industry actually running in a consulting form. So, we put all the development, academic research and key industrial players, big companies, SMEs together for the consulting. So, the real application programs can really feedback from the customer, how the developer and the technical research to revise the core and solve things again between research result and the industrial applications. So, we're very lucky. So, we did the two minutes work in the past two years. We have the ROS industry consulting Asia Pacific branch set up in ARTC. So, this is partnership with NTU. And we just will officially kick off this construction from next week. So, we will bring big companies, SMEs, and also developers together to promote ROS and ROS in the field application in the Asia Pacific area. So, just some successful story, how ROS is accelerating our innovation and research. The first picture probably most people know that this is dab challenging. So, like 16 members teams in the dab challenging, they're programming the subject in the humanoid of all the ROS. So, all the ROS can handle so many technical details. So, the millions of lines for the team control such a deep water complex humanoid with like 20-30 degrees freedom, the climbing layers, opening the widows and the so different disaster areas scenario. The second picture is the arms picking challenge. I was lucky to went to Seattle in 2015. Amazon host a robot picking challenge over there. We founded him in NTU. So, we participate in the challenge. They are like 20 teams over there. 80% of the users are using ROS to program the robot. So, this is combined with vision, 3D vision to perception and distinguished objects put into a shelf, you need to pick the right object and put into orbit. Motion planning has optimization. They give you like 15 minutes and the different objects have different points. Some are quite easy to pick. Some are very deformable, transparent object. Very challenging to perception and trust. So, the third one actually there are some very successful startups and companies using ROS as a software to figure out products. Because people always question are ROS stable or reliable enough to be implemented in the products. I think the answer now is yes. Because the best robotics, these people are a core member of a garage, one company and you make these two robots for logistical applications. Because the e-commerce company have a very strong demand for logistical robots in China and in the US. So, they are doing some testing in the warehouse. This one is a candidate for hotel service robot for debuting objects from the reception area to the door. So, when the robot reach your door, automatically give you a call and you can open the door and pick up the objects. You don't need to go down, wait time, wait to pick up a simple thing. The robot can help you to do it. This is AMAP, actually this is a local startup company. When I was a new, we founded this company. So, we build a robot in our dumpery. So, it took like half a year to build a robot with only two engineers. So, now it's about success. We get a tremendous support from the government. We just receive the POV grant from spring. Also, we have seen the Enjoy Investor in local and the Sound Investor from China. So, the story is saying that the robot making robot is not suited but as we meet like five or ten years ago, you meet the group of people, you need to spend a lot of time to do it. This robot is most of the components of the shelf. All in itself, the way to make it by itself. So, we use loss to build the capability on the work of other people. It's very easy and quick to build. Like John said, he took one month to build a robot and I think it's possible. Right? So, actually this slide is talked about in the day-to-day style. The theme is from lab to market. So, they also like to share some stories. How, what's the gap? The one thing we need to consider about using loss is kind of open source software. The first issue is the licensing and the IP issues. There are many licensing in loss. The BSD license, MIG license, I have two licenses, maybe make you confusing. But if you use a back-to-all MIT license, you'll be very happy because it allows you to modify it. You can use the software even for commercial use. For example, your startup company, you make a product. Your software is using a back-to-all license, then it's okay. There are just some regulations that you cannot use each company name to promote your product. But if you have some packages, which license is GBL or AGLA, your lawyer will probably make you cry because this license, they say you need to open source your core work because you use the software. So you need to be very cautious when you choose to use other people's work, especially for commercial use purposes. Also, there are some IP issues you need to include in your license file before you make the open source or inventing a product. Probably other people will see you if you use an illegal license. About the production, since we have a product, we probably need to think about how to transfer this demo to a branch of products. How can you maintain all these things if you have 101,000 customers? Because the open source software has some disadvantages. Even packages relate to other packages, like an industry. So if other packages have some upgrade and other packages don't have, there will be some independence problem, and then it's very difficult to maintain this software package and make it reliable. But the good news is nowadays we have CI tools integration. So I would recommend everybody, when you're developing a growth package, where you start to use these tools, it makes you continuously update and harmlessly integrate with other packages. Of course, there are also some testing and production tools, like DOC. DOC is a container technology that you can package everything in a huge package. So this package can be implemented on different platforms, but there will be no problems. So this kind of cloud-based technology, you can settle that technology. Also you can use janking tools to automatically build your source code. For example, you can use janking with your GitHub account. So when you change some changes of your source code, this all automatically builds and tests it for your end user or for your customer. For people who want to do industry IoT, Ubuntu has a new version called Ubuntu Core. So all your software will be derived as a snappy. So it's not a debbie, binary. So it's a kind of snappy. They use container technologies like a ship that have a lot of containers. So there will be no conflict between different packages. You just do an upgrade. Otherwise, get away and your customer will keep updating. And if the update fails, you can roll back. So this is quite a reliable and easy to maintain upgrading customer care. So there are also new business models because nowadays the global... the intelligent hardware is slowly moving from hardware hardware by providing service content and even build an ecosystem. So if your customer cannot afford a robot, for example, it costs like 30,000 K, 40,000 K, the customer cannot afford it. You can do a decent model. So you provide a robot and other maintenance service kind of service. So if you're doing pure software, you probably can consider about how to put your software in cloud. For example, in AMD, AWS, they have... So this software, the same actor, can be shared by hundreds or tens of thousands of customers with existing software API. So this is a kind of a business model. It's very easy for you to maintain one software than maintaining software practice installed on your product device. Also the pay-as-a-service of these kind of things. So all this technology, I think, is moving very fast. We're all talking about not robots anymore. Nowadays we are talking about cloud connecting robots. The software will be running in the cloud. The robot itself can be very cheaper in the future because it doesn't have so much high competition over all these high-level programming on the cloud processing or machine learning can be done in the cloud and can only be done in the cloud because as a Mac or developer, you probably can do some demo about machine learning, but you cannot surpass what's again the deep, like Google Deep or Amazon Deep. You probably need to use the API to process one image to run your robot and then feedback to come and control the robot. So where you might... So if you have some application, you probably need to think about how to use ArtCube's work or how to find a better way to deploy it and send it to your customer. So I think last, I'll just do some setting for an event. So next week we have a second-large industrial Asia Pacific workshop, which is hosted by ARPC. So we're called for registration now. I'm bringing our hearts, bringing some players. This year we have invited a lot of very nice speakers from US, from ROSE, or RSI, or ROS, Southwest Research Institute, is an American Association for Rosemonstryal. We also have some important speakers from NASA about the robot, and also some robot companies from Denmark and Europe. So there is a very standard art information service session. And also in the ARPC we have a very nice... We range from more than 10 very nice demo about the robotic process and also ROS and ROS industrial applications. So I encourage you guys to do the registration on the high-stage this event. I think this is my contact information, so if you have any questions, any interest about the ROS, ROS industrial, please contact with me. Yes, thanks. Any questions? How's ROS adapting on ROS 2.0? I mean, are there any developments at the moment? ROS 2.0, I have attended the two sessions last year in Seoul about the ROS comp. So they mentioned about the ROS 2.0, so we will release a better version in Q3 or Q4. But many works are still in the ROS 1.0, so ROS 2.0. But we keep watching it. So also for us it's a learning journey because ROS 2.0 has a major revision because it's based on BPS and there will be no ROS core. It can be also running embedded system for real-time applications. So this is actually the good news for ROS industry because for industrial applications sometimes we really need real-time control of the motors and the robots. So we have people watching it, but I don't know, maybe the people from ROS 2.0? Yes, they will share this information. Any other questions? Any more questions for Dr. Liang? Okay, thank you. Thank you Dr. Liang.