The poster session of the DLRV (Deep Learning for Robotic Vision) workshop will be from 10:45 to 12:00 on 21 July, 2017 at Hawaii Convention Center in Honolulu, Hawaii, USA.
It’s a crazy time of the year! We have more visitors and media interested in visiting Robotics@QUT than we can handle, we have ACRA (the Australian robotics conference) and ICRA (the biggest robotics conference in the world) submission deadlines, we have Robotronica, QLD State Advance grants due, a new teaching curriculum course rolling out, and on top of all that we need to continue our normal research and teaching duties. How do we deal with all this work and achieve a reasonable work-life balance?
For a while now, I’ve noticed that we’re all super busy. I often ask myself why am I busy and how can I be more efficient at work and at home while remaining productive. Michael Milford gave a great talk at the recent Robovis on work-life balance highlighting that you have to put in the hours at certain times of your career. This doesn’t mean you always have to work all hours of the day all the time but coming up to a deadline, this might be required.
I think that we choose a career in research and teaching because we love the work. It doesn’t pay what industry would but it’s still fairly comfortable. And you have a choice in where your career takes you… this freedom to me is very important
We love our work and because of this, we often think about our work outside of standard 9-5 hours. On the flip side, do we have to work 18 hour days and weekends? Hopefully not for an extended amount of time. We should have time to wind down. I think the following article really captures how I think of work and life: https://hbr.org/2015/04/its-the-weekend-why-are-you-working? “Our passion for our work and the pleasure we gain from feeling productive may explain why we so often work on the weekend, but we still need to be sure to make time to recharge”.
The moral of the story is that if you are busting a gut now, make sure that you leave some relaxation time after your deadline to recover. If you are not leaving recovery time for yourself, you’re not working at your optimum… give yourself a break when needed.
For those in the deep dark tunnelof writing a submission to ACRA and ICRA, please know that there is a light at the end of the tunnel. These are tough weeks but soooo worth it. Getting a paper in ACRA is a great chance to speak to the Australian robotics community while getting an ICRA paper is one of the academic pinnacles in publications for robotics.
I want to take this chance to thank everyone who has put in a huge effort with Robotronica… it looks like it will be an amazing day because of these people. Check out the sheer number of events on the Robotronica site: http://www.robotronica.qut.edu.au/ . We’re still looking for volunteers so if you have spare time, please make yourself known if you can help out.
Visitors and Potential Industry Partners
Thank you to all those who have put in huge amounts of time for demos and presentations to various visitors and potential industry partners. The payback may not be immediately obvious to you but Robotics@QUT as a whole benefits immeasurably from these professional demonstrations. As a result, we get more people at our door who will fund all the fun projects that we’re involved in. Indirectly, this means the group as a whole stays healthy in terms of funding and culture.
Finally, during these hectic weeks, I want to thank you all for such tireless enthusiasm and effort. We all really appreciate it.
I’ve been thinking a lot about where RAS is going at the moment. Individual groups/centres are doing fantastic things but as a Discipline we don’t have a common direction and I think we have the expertise to do really cool things that aren’t feasible as individual groups. I’ve been struggling to make our vision more focussed as our group is so diverse and does so many different things on different platforms. The only common thing is that we want to impact society through our research, our gadgets, and our education… it seems that what we all have in common is that we want to build/create/provide robots for the real world.
I believe the other thing that we have in common is the requirement for a set of platforms that are not only easily configurable at the software level but also the hardware level i.e. a common customisable robot platform that aerial, ground, and underwater roboticists can rapidly develop to a working system both for research, industry, proof of concept widgets, and education.
I think as roboticists, we use the argument that robots are really hard to make and due to the diversity of the robots that we work on requires us to be artisans of a particular technology, algorithm, widget, etc. However, I think this is part of the problem… we’re stuck in the artisan era where we build beautiful robots or components of robots and can only do this rarely. This results in very expensive (both in time and money) platforms/software/widgets.
The next step from being an artisan is mass production… i.e. automate the process to make exactly the same thing at high efficiency enabling the cost to be significantly reduced. Some robotics companies have tried to do this and I think UAV/drone companies are starting to make it a successful enterprise.
Outside the robotics domain, IKEA has taken the mass production one step further where the customer receives a flatpacked product which the customers have to assemble. I think Ray Russell and Owen Bawden have taken this step in robotics with AgBot 2. This platform can be sent to the customer who assembles it in their garage. Daniela Rus from MIT has also shown these self-assembling robots which have a very similar idea behind them.
I think this idea hasn’t taken off so much in robotics development as roboticists want/need to tweak both hardware and software. Different modules are always required depending on the diverse tasks that robotics can be employed for. For example, on a UAV I need lightweight payloads where as a self-driving car does not have this constraint but might need high end computing systems.
I wonder if the answer lies in the idea of mass customisation. With the development tech that we have at our disposal now such as 3d printing, laser cutting, etc., mass customisation is a reality. IKEA has made forays into this with the ability to design your own kitchen and someone comes out and installs it once you have decided on the kitchen best for you. Prosthetics start up companies are exploiting this as everyone’s body shape is different. Even Google is getting into it so that people can customise the phone they want. Project Ara is driving this with Google’s support and have called their phone platform an endoskeleton. The endoskeleton is a bare bones frame with comms and power in-built. You can add your own modules and develop your own modules to provide a customised phone.Furthermore, Daniel Rus showed similar work with the ability to customise your robot which is then 3D printed and loaded with the right software to run the robot.
What if we did same thing, not at the toy robot level, but at the “real world” robot level? For example, we could build a robot platform endoskeleton that I could attach my sensor modules, my compute modules, my actuation modules, etc. We could have a couple of endoskeletons (small and big) so that we could accommodate a variety of platforms such as UAVs and ground vehicles. Sensor and computer modules would be interchangeable for the platforms but the actuation modules might be only created for specific purposes such as wheels vs a robotic arm.
What I really think is that the tool is not the key i.e. the endoskeleton will not solve all our problems. We need the right mindset and people who can do this rapid development by creating the right tools for the job i.e. create robots for the real world – for research, industry, and education.
Why I think this is a good idea
The idea accommodates everyone’s interests in the Discipline. For example, people working on UAVs for infrastructure monitoring can contribute new hardware modules for aerial problems such as biosensing payloads while people working on manipulator design can make an arm that could be attached to a pioneer or AgBot 2. In effect, we’d be doing what ROS did for robotics software, for robotics hardware. Furthermore, by opening up the module interface development to the robotics community, new hardware modules could be produced by groups with complementary expertise… for example a super efficient battery system. We can also develop software systems that make the decision on best systems for purpose through decision theory.
One of the great things about the development of an endoskeleton is that it’s impact on the research projects we are already involved in is useful for these projects as well. If we begin to all work on a modular endoskeleton and modules for our individual projects, we would be contributing to the overall goal of mass customised robotics systems anyway. In this way, we don’t need to take resources away from existing projects but could start thinking about accommodating the idea of mass customisation.
Finally, I think if we had the ability to turn around a prototype robot by plugging a few modules together, we’d be agile enough to address industry requests for quick research ideas that they may want to test. This allows us to build up strong relationships quickly rather than waiting for ARC linkage funding to build a platform for example. This is very much in line with QUT’s blueprint and vision to build a name “for the real world”. The ability to turn around a platform quickly also allows us to address our own vision of impacting society with real world widgets. We no longer have to wait for a 3 year development of a robot before testing our ideas… we could make a platform within weeks!
This is a very nascent idea and I’d love to hear your feedback on feasibility, excitement about the potential direction, how it does/doesn’t fit into your own group’s/centre’s directions, etc. I’ve also provided some links below about the phone endoskeleton supported by Google.
Imagine phoneblocks for robotics:
Phoneblocks is now a real thing called Project Ara which google is supporting:
I was at the International Conference on Robotics and Automation (ICRA) last week. This is the largest international robotics conference of the year and had 3000 people attending throughout the week. There were many comments about the size of the Robotics@QUT contingent which is great! We had three workshops that our group organised: Peter Corke’s Challenges and Opportunities in Robotic Vision, Niko Sunderhauf’s Visual Place Recognition in Changing Environments, and Robotics in Agriculture which David Ball and I organised. All were very successful. I believe that there were over 400 people (see the image below) at Peter’s workshop which is amazing. We were invited to create a special issue in the Journal of Field Robotics for the Robotics in Agriculture Workshop. I also highlighted in the Google+ robotvision community, plenary talks that I liked. These talks were recorded and posted on youtube.
The conference is a great opportunity to see the latest work in the field but also an excellent chance to network with people in the robotics community. I’m still learning the importance of networking but many collaborative opportunities are opened up to you when talking to people over a coffee. You can also learn a lot about what the next big thing is going to be by talking to robotics groups not so geographically isolated as us Aussies. I know there’s a bunch of conferences still to come and I encourage you to challenge yourself by talking to someone outside your group of mates… maybe just by telling a presenter that you really enjoyed their presentation and work and would love to know more.
Another thing that was very prominent at the conference was the number of robotics companies hiring. It was like being with a bunch of sharks… all of them gobbling up people as quickly as they could. This is exciting times we’re in when there is so much buzz about robotics with companies clambering over each other to hire us roboticists! But it’s also a bit scary when companies are literally gutting academic groups (take CMU for example). These companies are very secretive about what they’re doing while at a conference designed for open presentation of our work and research… a little a bit contradictory. I think there needs to be a balance between the give and take from the different communities. I’d like to see the robotics companies giving back to the conference in some way rather than just taking.
On a note closer to home, Insight Survey for undergraduate unit feedback is out. Please encourage your students to respond to the survey and get the numbers up. Did you know you can tell the number of responses through Business Objects via QUT Digital Workplace? It’s hard to find but go into Evaluation and then search for your unit. I’m up to about 30% response for both my classes and am aiming for 40%. Let’s see if we can get those percentages up for the whole discipline. Good luck everyone with teaching surveys and the end of the semester.
I want to be an astronaut. We had a talk yesterday from Professor John Norgard who works on electromagnetic interference testing for NASA spaceships. It was a really interesting talk on the various Space Shuttle and International Space Station missions that have occurred over the last 30 years. He then continued to talk about the testing of the Orion Spacecraft slated for manned missions to Mars. It was like watching the scenes out of a movie with Plasma drives designed for propulsion, new parachute deployments required for reentry, missions to the Moon, missions to asteroids, and missions to Mars with a flight time of 3 weeks! It was very exciting to see and hear about the exploits of NASA.
I have only been involved in one space project where we automatically tracked the reentry of the spacecraft Hayabusa, to Earth ( a film of the reentry is here https://www.youtube.com/watch?v=ZCdGQhYLo30). This spacecraft landed on an asteroid, collected some dust by shooting a bullet down a tube at the asteroid, flew back using Ion drives which all failed. It managed to limp home after a 7 year round trip, landing in the desert of South Australia with a payload of micrograms of asteroid dust (http://global.jaxa.jp/projects/sat/muses_c/). It seems that many space missions and the story behind them are like a movie and have an amazing impact on what we know about the world and our universe.
Before this talk, my impression was that NASA had lost it's drive for new and amazing things... how wrong I was. The talk prompted me to look at NASA's Vision to remind myself about impactful science and what this massive group has achieved and wants to achieve. Their vision is "To reach for new heights and reveal the unknown so that what we do and learn will benefit all humankind." I think this vision statement really suits NASA.
I've been peripherally involved in the new Vision for SEF which I mentioned briefly in my last post. One of the statements about the current SEF Vision draft is that it doesn't differentiate our Faculty from any other EECS-like faculty... what makes our faculty different from the thousands of others? You could almost say the same about NASA's vision statement but I still like it once you put it in context.
These questions prompted me to consider RAS's Vision statement "To create intelligent robotic and autonomous systems that benefit the world". Does this differentiate us from any other robotics group in the world? What makes us different? I see a talk about NASA and I want to become an astronaut at NASA... what makes people want to become a roboticist at QUT? What makes us unique? We have an amazing group here at QUT and I'd love to capture the great things we do and are aiming for in our Vision. Can I get everyone to have a think about this as we'll revisit our Vision in the not too distant future.
This is my first blog ever... as I don't often have much to add to all the information that is already out there in the ether. However, I don't get to speak to you all as often as I'd like and thought that this might be one way I can convey information related to Robotics@QUT a bit more regularly. So please bear with me as I get used to writing a blog. Any feedback/comments are more than welcome.
I've called this blog "The Weeks that Was" as it was originally going to be a weekly thing but it'll probably be more likely fortnightly. So what happened over the last couple of weeks? I think the major thing that happened to all of us was the move from S to G Block. I want to thank everyone for being so patient and accommodating with the move. It seemed to be fairly seamless... except when I realised that I hadn't organised for the equipment in the Fishbowl Lab to be moved on the Thursday when we were supposed to be finishing packing... minor panic. Thank you everyone for pitching in to pack the lab at such late notice. I don't think G Block is as bad as we all thought it was going to be. A bit of nice furniture in there went a long way to sprucing up the place. I think hanging out in the cafe is also a big plus. Please let me know if you have any problems and we'll do our best to sort them out.
Everyone who was teaching undergraduate units this semester received their Pulse Survey. This is the main form of feedback from students at the half way point in the Semester. I'm fortunate enough to be working with a fantastic teaching team (Lawrence Buckingham, Ruth Schulz, and Ben Talbot) who have put in an amazing amount of work into CAB202 (C Programming and Microcontrollers for 1st and 2nd year students) which needed an injection of energy. I'm pleased to say that we are looking much better in this Unit than we did last year. I hope everyone else has had positive feedback from their Pulse and if not, please come and chat and we can start figuring out a way to improve your scores.
Last Monday I attended the SEF Wider Leadership Group Meeting which was chaired by the Executive Dean, Gordon Wyeth. Participants in the meeting included Discipline Leaders, Heads of School, Portfolio Leaders, and Assistant Deans. The purpose of the meeting was to provide feedback on a very recent draft of a Vision for SEF, put together by the SEF Executive. As it wasn't a final version and requires feedback from a number of stakeholders, it will go through significant refinement over the coming weeks. I won't go into detail here due the potential changes it may go through but I thought I'd highlight the common thread that I picked up on - the development of lifelong relationships... with students (from early education to alumni) and industry (partnerships rather than contracts/transactions). I found this was fairly powerful and really suited the culture of QUT.
Finally, I'd like to end on PPRs (Professional Performance Review (and Plan)). All staff have to do a PPR each year. This begins with a review of last year and then ends with a plan for the following year. Most staff in RAS should have completed their Review (if not the Plan). I find PPRs inspirational to be a part of... understanding the goals and directions that people are heading towards and seeing them reach and surpass those goals is a really positive experience. I really encourage everyone to think about their PPR plans... what's the thing that gets you up in the morning? Make sure you give yourself time to work on that thing!
I hope this blog wasn't too long and is somewehat informative. Again, any feedback/comments are very welcome.
I'll be back on here next week(s).
Comments can be added to the page Duncan created at HDR Processes and Student Experience
QUT's team, a team of undergraduates led by Matthew Dunbabin and with Bruce the boat, came overall third in the first ever RobotX maritime challenge.
Final Standings from Singapore were:
5. National University of Singapore
6. Osaka University
There were three Australian teams, none of the others placed.
The QUT team also won:
- Best Journal Paper (tied with MIT/Olin)
- Environment Award.
The QUT robotics group has been busy publishing papers across a wide range of conferences and journals over the past few years. Below is an animation of the paper wall from a couple of years ago to the present day, just after the submission deadline for the 2015 IEEE International Conference on Robotics and Automation.
- Scour through old datasets in the hopes of finding one with shots of the doorway
- Take a current photo of all the papers
- Run our alignment algorithms
- Add labels, re-animate and export to gif
Multiple PhD positions are available across the Robotics and Autonomous Systems discipline at QUT.
Consult the list of projects and the process for applying on this page: http://tinyurl.com/mve7fnq.
Note that the deadline to apply for a scholarship that starts in 2015 is 30 September 2014.
The lab is organising the 1st QUT Winter School on Robotics and Autonomous Systems on 1-2 July at the Gardens Point Campus in Brisbane.
Undergraduate and Masters students from all over Australia will learn about the latest in robotics research and how to contribute in the future. The event will feature talks by QUT academics as well as 2 prestigious invited speakers from UQ and the University of Sydney.
For more information, see http://tinyurl.com/QUTwinterschool
The CyPhy lab has had 3 papers accepted to the 2014 IEEE/RSJ International Conference on Intelligent Robots and Systems to be held in Chicago in September, including collaborative research with the University of Antwerp.
This just arrived. The plan is to attach it to the Guiabot so it can open doors, push lift buttons and generally cause mischief. Come by my office if you want to touch it.
I guess the converse is important as well...