locked
Academic acceptance of MSRS RRS feed

  • Question

  • When perusing many online articles about robotics, it's mentioned repeatedly how the academic community has not embraced MSRS and how the robotics manufacturing world has indeed embraced MSRS.

    Why is the academic community reluctant to embrace MSRS? Does anyone have a clue?

     

    Sam

    Friday, March 28, 2008 4:48 AM

Answers

  • My two cents worth is as follows. This is all my personal opinion.

     

    Based on the number of e-mails that I get from students, and postings in this forum, there are quite a few institutions using MSRS for semester projects and even thesis projects. So academia has picked it up to some extent. However, I will admit that uptake has been slow.

     

    Also, following on from what Ben said, my experience is that there is a large amount of code out there for Unix-based systems but that the authors are reluctant to port it to Windows. And why should they? There is nothing to be gained if you already have a working system and you are trying to finish your PhD. On top of that the learning curve for MSRS is steep. I've been trying to assist with that by writing tutorials and now a book, but it all takes time.

     

    As I said to somebody earlier today, if we want a bridge then it will have to be built from the Windows side across to Unix. I actually think this is a worthwhile project because there is a wealth of code available, especially for things like Computer Vision, Mapping (including SLAM), Path Planning, etc. Furthermore, the small embedded systems like the Gumstix all run Linux, and it would definitely be good to integrate them into MSRS.

     

    I think that one reason MSRS uptake has been slow in the research environment is that so much stuff needs to be written if you are doing serious work like SLAM. Over time this will come from the MSRS community and maybe Microsoft and then there will be more incentive to use Windows.

     

    Anyone want to start a new Open Source project to migrate CARMEN to Windows for example? I'm happy to help and I have already made a start on it, but it's not a simple task. However, this raises the issue of maintenance -- how would the Windows version stay in synch with the Unix version?

     

    The Simulator in MSRS is great. Perhaps we can leverage that by writing code that generates log files in the CARMEN log file format, and using some CARMEN modules to draw maps, etc. From a teaching perspective, I think this would be great because it is zero cost to both the student and the institution, and lets them learn advanced robotics concepts. I'm not saying there is anything wrong with Player/Stage, but I think that the MSRS Simulator is better :-)

     

    Trevor

     

    Saturday, March 29, 2008 12:54 AM

All replies

  •  

    In my case I’m learning MSRS for academic/research applications and I’ve got approval from my university and supervisors to utilize it in my Master Thesis.

    Maybe it just need more time to be spread more in the academic communities.

    Friday, March 28, 2008 11:24 AM
  • I think a common case is that many established robotics researchers already have mountains of code for robot architectures, behavior, computer vision, etc. and it is all based on Linux. 
    Friday, March 28, 2008 2:24 PM
  • We've been using MSRS in our academic lab for over a year now. There's information about the lab in the latest issue of Robot magazine:

    http://botmag.com/articles/CoroWare.shtml
    Friday, March 28, 2008 7:04 PM
  • My two cents worth is as follows. This is all my personal opinion.

     

    Based on the number of e-mails that I get from students, and postings in this forum, there are quite a few institutions using MSRS for semester projects and even thesis projects. So academia has picked it up to some extent. However, I will admit that uptake has been slow.

     

    Also, following on from what Ben said, my experience is that there is a large amount of code out there for Unix-based systems but that the authors are reluctant to port it to Windows. And why should they? There is nothing to be gained if you already have a working system and you are trying to finish your PhD. On top of that the learning curve for MSRS is steep. I've been trying to assist with that by writing tutorials and now a book, but it all takes time.

     

    As I said to somebody earlier today, if we want a bridge then it will have to be built from the Windows side across to Unix. I actually think this is a worthwhile project because there is a wealth of code available, especially for things like Computer Vision, Mapping (including SLAM), Path Planning, etc. Furthermore, the small embedded systems like the Gumstix all run Linux, and it would definitely be good to integrate them into MSRS.

     

    I think that one reason MSRS uptake has been slow in the research environment is that so much stuff needs to be written if you are doing serious work like SLAM. Over time this will come from the MSRS community and maybe Microsoft and then there will be more incentive to use Windows.

     

    Anyone want to start a new Open Source project to migrate CARMEN to Windows for example? I'm happy to help and I have already made a start on it, but it's not a simple task. However, this raises the issue of maintenance -- how would the Windows version stay in synch with the Unix version?

     

    The Simulator in MSRS is great. Perhaps we can leverage that by writing code that generates log files in the CARMEN log file format, and using some CARMEN modules to draw maps, etc. From a teaching perspective, I think this would be great because it is zero cost to both the student and the institution, and lets them learn advanced robotics concepts. I'm not saying there is anything wrong with Player/Stage, but I think that the MSRS Simulator is better :-)

     

    Trevor

     

    Saturday, March 29, 2008 12:54 AM
  • I would like to light up the topic from another point of view.

    I am a PhD student, who makes research in interaction control of robotic arms attached onto mobile platforms. On interaction I mean the situations when the robot get in contact with its environment, and the control strategy has to consider the dynamics of the complex system (compliance control, self-collision avoidance etc..).

     

    It would be very grateful to find a simulation environment, which is accurate enough, free (at least for academic purpose) and easy to use.

    In the past few weeks, I made a trial with MSRS simulation environment, and I was surprised that my custom platform and articulated arm how easily managed to be implemented (thanks for many experts on this forum).

    After the happy beginning, I have met some issues of the Ageia PhysX engine like the wobbling joints and cheating with joint projection. I have recognized that the Ageia engine is not for scientific applications if the accuracy of simulated interaction forces is very important.

     

    However, the simulator is extremely useful for researchers who mostly interested in SLAM and other algorithms in which the dynamics is not particularly important.

    (I am not an expert of localization, but I guess that grabbing of lifelike odometry data from the simulator is much easier than to retrieve contact forces between the actors).

     

    I am not sure if my opinion is correct or not, because I am new to MSRS. So if someone has different experience in the before mentioned application of MSRS simulation environment please contact me or just post to here.

     

    Monday, May 19, 2008 9:13 AM
  • As a beginner and student, I see lack of availability of literature/ books is the most important hurdle.

    kr

    Thursday, May 22, 2008 3:04 PM
  • Sara Morgan already has a book out that is published by Microsoft Press.

     

    Kyle Johns and I have another book coming out right now. It has started shipping and should be in the bookstores in the next couple of weeks.

     

    Go to Amazon and search for "Microsoft robotics" and you will find both of them.

     

    Obviously, these will not be the last books on the subject.

     

    Make sure that you do all of the tutorials, and also look at the Introductory Courseware. There is a heap of information on the Microsoft web site in various online user guides. Of course, I understand that a book is preferable to online documentation because you can read it on the bus to work/school and take it to bed at night if you can't sleep ;-)

     

    Trevor

     

     

    Friday, May 23, 2008 12:27 AM
  • I am completely new to robotics. Even though I am a windows developer, I have difficulty in understanding various concepts, terminology etc related to this technology. I am getting better, slowly.

    Yes, I have been reading posts in these forums and know about these two books. I have already ordered the book by Sara Morgan and waiting for the availability of your book.

    Tutorials and other reference material is good and I am going through it but a good book on the subject is way different. And one or two books on the subject do not give you much choice.

    A while back when I told my advisor at the university about this technology, he was unaware of its existence. He ordered the book on my request. There is another group of students under his guidance who is doing a masters level project about robotics and they were too unaware of it. If our university library had many books on the subject, the awareness of this technology within our department may have been much better. This was the reason I posted my earlier comments.

    By the way, Trevor, sometimes I wonder how do you get time to respond to so many posts. You are doing a great job and I know that I have learned a lot just by reading your posts. Thank you very much.

    (I hope I will be able to take these books to bed and not go to sleep immediately after ).

     

    kr
    Friday, May 23, 2008 5:25 PM
  • Another big problem with the slow academic acceptance issue is expectations.  I don't think that Microsoft Robotics Studio meets expectations.  Perhaps everyone has heard of the acronyms Commercial-off-the-shelf (COTS) software vs Research-off-the-shelf (ROTS) software.  Microsoft Robotics Studio is research quality software, not commercial, but nobody knows that until they try it out.  The few robotics researchers who are inclined to try something from Microsoft are confronted by sketchy or out-of-date documentation, a minimum of examples, a steep learning curve, and significant bugs that are still being fixed with each release.  Yes, they can do real research in robotics studio, but it isn't what you would expect from a commerical product.  After being disappointed in the beginning, how many give up too early and turn to other robotics control platforms such as Player (for linux).  Arguably, Player isn't any better... but it isn't any worse either, at least as far as research is concerned.

     

    This slow pickup of Microsoft Robotics Studio is a real shame, because I suspect the developers are caught in the middle.  While academics and hobbyists clamor for help and documentation, the staff will probably remain under-funded and under-staffed until such a time as Microsoft knows what to do with its Robotics Studio.  Another forum post asked the question, what does Microsoft intend with this release of Robotics Studio?  Who are they targeting?  It's just a guess, but I suspect Microsoft released the robotics studio only to position itself for the likely commercial robotics explosion.  When there is money to be made, Microsoft wants to tell all the potential customers that they have been in the robotics business for a long time with their Robotics Studio.  At such a time, Robotics Studio will no longer be free, and we will see Microsoft put real money into the project. 

    Saturday, May 24, 2008 10:01 AM
  • Hi Khalique,

     

    Thanks for the kind words.

     

    I appreciate that documentation is a big hurdle and I agree that more books will help. I don't even mind the competition :-)

     

    I plan to work on more books, but it took six months to put our first one together so it is not something that will happen overnight.

     

    As for the forum, I am currently self-employed so it's my own time. I check the forum most days, except when I am away like last weekend when I was running a Scout camp for 80 Scouts.

     

    Trevor

     

    Monday, May 26, 2008 12:24 AM
  • Hi Eric-B,

     

    There's an old saying: You can tell the pioneers - they are the ones face down in the mud with arrows in their backs.

     

    Robotics, even though it has been around for many years, is still a field in its infancy. Microsoft started from nothing about two years ago and I would be the first to agree that they still have a long way to go. However, it has to start somewhere.

     

    Remember that the Robotics Team is only a small group - about a dozen people. I think that they have achieved a lot in such a short time. Bear in mind that this is to some extent an experiment for Microsoft. The team is treadding new ground.

     

    If you want to look at it from a commercial perspective, the Robotics group was "incubated" inside the Microsoft Research group for the first year or so. In that sense you might call it research software. Now the group is trying to transition to the commercial world and that will mean some changes. I think you will see some exciting things coming in the next year or so.

     

    In the meantime, welcome to the world of the early adopter. I installed the first CTP back in July 2006 and I there are still some areas of the product that I don't know very well. What I can say is that it does get easier the more you use it. And the more people who use it, the more examples there will be. Eventually there should be a critical mass of people and a large, active community. But there will always be newbies. That's why I try to help, particularly on the documentation. (That's the teacher in me coming out).

     

    Anyway, that's my opinion and I admit that I am biased :-)

     

    Trevor

     

    Monday, May 26, 2008 12:35 AM
  • hi to everyone. i am academician working currently in robotics. i agree to the point of MSRS not meeting expectations. People are reluctant to accept new technologies unless they are able to use it for everyday practical purposes.and they see robotics either for toys or for industrial applications only. In the academe, in my opinion, robotics as a discipline is caught in the middle of tug of war of whether it can be considered as solely engineering or computer science field. i think its a merging of both. And there are limited courses that offer robotics as a program course. I am currently in the process of looking for ways to come up with a curriculum for robotics. I must admit though, that i am a newbie in MSRS. But i see it a s good platform to bridge the gap between engineering and computer sciences. To this, i am asking for any help that can be given to me by the good hearted and highly qualified people here in this forum. Any help will be appreciated. Thank you in advance
    Tuesday, June 17, 2008 1:22 AM
  • I just started on my Master's Thesis, and the professor let me choose my environment for developing a swarm robotics simulation.  It came down to Webots vs MRDS and I picked MRDS for various reasons.  I believe you will see MRDS popularity grow quickly in academia in the near future.

    Friday, June 27, 2008 12:16 AM
  • Hi,

     

    Ross de Rango at Deakin University did some work with multiple robots in the simulator some time back. I don't know if he is still working on it.

     

    You might also want to search this forum for "swarm" or similar words and also for "Ross".

     

    I just wanted to confirm that swarms are indeed possible in the simulator. However, for large numbers of robots you should probably create them in your code and not using a manifest.

     

    Trevor

     

    Saturday, June 28, 2008 12:59 AM