locked
The Case for Investing in a Robotics Studio 2011 RRS feed

  • General discussion

  • To Whom it May Concern at Microsoft,

    I'm begging to feel that Microsoft has allowed MSRDS to die off by failing to put serious funding behind this project and I fear that any day now will be the one when this product is formally announced as discontinued.

    With that said, I think there is a business case for MSRDS now more than ever and here's why:

    1.) The competition is investing in robotics, take Google's self driving cars for instance.

    2.) Microsoft has recently developed or announced plans to develop at least two great products that would integrate extremely well with MSRDS.  There is opportunity for cross pollination and leveraging of resources already at work with the following examples:

    -Kinect has already begun to show it's potential with hackers and alternate uses outside of XBox 360 gaming.  This sensor is a potential SLAM enabler with it's real time 3D imaging capability.  Robotics developers have been waiting for years for a near turn key SLAM navigation system, with the right software, this is possible.

    -Windows on ARM should at the very least make porting MSRDS or at least a runtime engine for MSRDS "apps" if you want to call them that on ARM architecture much easier.  While a substantial portion of mobile robots run on ARM cpus or microcontrollers; MSRDS has always had to run on x86 hardware.  So one would need a control PC remotely operating their autonomous ARM robots which is very inefficient.  It makes sense for real time autonomous apps to run directly on the target robot even if one wishes to retain a remote control PC for override or monitoring capabilities.

    For what it's worth, I hope Microsoft can share these suggestions up the management chain for consideration.

    Thank you

    Friday, January 21, 2011 2:55 AM

All replies

  • Thank you for your feedback. I willl bring this post to the attention of our General Manager.

    **PERSONAL OPINION ALERT -- THE FOLLOWING IS NOT AN OFFICIAL MICROSOFT STATEMENT**

    Let me say first that rumours of the demise of the Microsoft Robotics Group are unfounded. We are still alive and well and even growing :-)

    The value of Kinect as a robotics sensor is well understood by our team and we are investigating it but nothing happens overnight.

    Windows on ARM is still Windows so naturally we expect RDS to be available. Although an ARM SOC system running Windows might draw less power than an equivalent x86-based system and maybe even be faster, it is not the same as many of the ARM-based robots that are on the market today. Many of the small ARM boards currently on the market are incapable of running Windows 8.

    To create really intelligent robots you need more than a microcontroller with a small amount of EE-ROM and RAM. You need gigabytes of memory, WiFi connectivity and complex software for speech recognition, computer vision, SLAM, etc. Such systems need to run on a sophisticated operating system and require the equivalent of PC hardware. So my caution is: Don't assume that because something is labelled "ARM" that it will be able to run Windows 8.

    I think this is an exciting time for robotics. Over the next few years we are going to see a revolution in the capabilities of robots coupled with enormous price reductions and our team intends to be part of that revolution.

    Trevor

     

    Saturday, January 22, 2011 8:36 PM
  • Being an old programmer (Yes, there is gray in my beard!), I have seen this syndrome before:  "No major releases must mean the product is dead."  With the fast pace that many companies have of 'major' releases often leaves observers with this impression.  I understand this position and fall victim to it from time to time.  I (personally) find lack of releases a feature for a stable product.  (Is RDS bug free? No.  There have been 3 'service pack releases. Is the framework stable?  Yes.)  Being that these guys are developing the platform and not the hardware, I would expect releases only when the framework is changed or major new features added.  Personally, I'm hoping that there will pieces to allow higher level concepts such as vision to be used without having a PhD. to know how to use them.  (I can dream can't I? :>)  I see RDS in the same family as Dev Studio - two or three years between major releases is typical.

    I personally don't see evidence that the RDS group is dying.  I see research papers being published.  I see universities working with RDS to push the boundaries of what computers can control.  I see companies posting questions on usage.  (It's nice to know RDS is expanding.)

    Personally, I see way to much evidence that MS is doing their best to show off the 'eye-candy' technologies.  (How much of last year's PDC was devoted to the latest, newest GUI platform versus core OS/hardware issues?)  Face it - Robotics is only 'sexy' if you have a robot flailing it's arms shouting "Danger Will Robinson"!  Having an autonomous car is great, but a typical response from most people (software developers included) is "I can drive a car already.  Why is it so hard for a computer and why should I trust it?"  (Ever get bad advice from you GPS on where to turn?)  Having a Mars rover that is out-run by a tortoise is boring even if the technology behind it is ground-breaking.

    Just had to add my own two cents having seen this topic too many times.  Consider this a 'vote of confidence' in the progress and handling (so far) of the RDS group.

    Ed Harfmann

    Monday, January 24, 2011 5:54 PM
  • Thank you for your passion for RDS and your suggestions.
     
    Microsoft is very excited about the Robotic opportunity and is making a substantial investment in this space.
     
    So why haven’t you heard any substantial announcements about our roadmap lately?
     
    After the release of RDS 2008 we came to the conclusion that providing development tools to the community is not by itself sufficient to kick start Robotics market that  is relevant to most of us. By no means I don’t want to belittle Research’s exciting work, the continuous advancement in industrial / service robots, and the growing Robotics enthusiasts that have hadmajor impact within education. However since the early 2000s we haven’t seen any products that a consumer would point out as having direct impact in their day to day life.
     
    Why?
     
    We identified 3 core areas that we need to address:
     
    • Technology (T):  There are some fundamental and hard problems that research has been tackling for decades. How do we develop an effective Human Robot Interaction model? How do we navigate an unstructured environment in acontextually aware fashion? These and others are questions that we need to answer for a hardware platform that is accessible to a broader audience. Existing solutions either require expensive hardware solutions or haven’t reached to the fidelity and reliability required for consumers.
    • Experience (X): What are the scenarios that are compelling for consumers given the limitations of the technology? How do we create experiences that are different from other mobile devices (e.g. laptops, phones) and still delight consumers and add value? In my opinion, this remains unexplored territory.
    • Business (B): For Robotics to become mainstream, compelling and sustainable business opportunity for an ecosystem of developers, hardware manufacturers, and other professionals involved in the product pipeline is requiredThe Robotics industry will need to look to other industries and market segments to help with commoditizing component costs with their high volumes and lower prices which will help fuel the Robotics industry.
     
    Microsoft continues to invest in Robotics while championing the need to address all 3 core areas.  As we continue to push forward on these challenges, I appreciate your patience. We are working hard not to disappoint you…
     
    Stathis

    Stathis Papaefstathiou | General Manager | Microsoft Robotics
    Wednesday, January 26, 2011 6:25 AM
  • "After the release of RDS 2008 we came to the conclusion that providing development tools to the community is not by itself sufficient to kick start Robotics market that  is relevant to most of us"

     

    well said.  it would have never been enough when windows was first being developed, it certainly is not enough now. instead it takes vision and long term commitment, something long gone from Microsoft, but bound to come long in the next big person/company.

    Wednesday, January 26, 2011 4:05 PM
  •  

    Stathis,

    Great post and I'd like to reply directly to a few of your points.

     

    "How do we navigate an unstructured environment in acontextually aware fashion?"

    One of the problems I've raised before is that we keep solving the same problems over and over again.  The development platform needs to consolidate and provide all of these solutions in a unified environment so that commercial developers don't need to reinvent SLAM or machine vision algorithms for instance.  You need to take technology 90%+ of the way for the commercial sector to bite, especially in today's economy.

    The consumer and industrial product industries don't like to invest more than 12-18 months on developing a new product (not to be confused with platform development, early stage research and derisking activities) and thus at that point there is no longer time for science only engineering.  In other words these companies, especially small entrepreneurial ventures will package available technologies in a product design cycle that they can source.  There is little to no opportunity at that point to develop new technologies.

    If one could build a robot out of supported components and have MSRDS configure it automatically as Windows does with PnP in Windows that's a start.  Second, I feel MSRDS needs to roll in technological solutions as they mature.  I'd argue that there are plenty of robust SLAM algorithms out there that could be refined and incorporated into MSRDS, but it is key that the MRDS GUI provide powerful yet simple to use tools for utilizing SLAM.  As I've suggested before this should be full of wizards and such for configuration and not require a PhD thesis in itself.  Just look at the amount of complex image manipulation one can do in Photoshop without writing any code.

    Again the PC never took off until it shipped with an OS and applications out of the box.  Before then, one had to start from scratch and write an OS and program just to use it, even if others had already done so.

    MSRDS needs to serve the needs of potential robot developers by providing solutions to the problems that have already been solved such that developers are building on top of prior work rather than starting from scratch each time.  There are other robotics projects out there that share this philosophy, take ROS.org for instance.  However ROS doesn't really compete with what Microsoft is aiming to provide and that's an easy to use development environment.

    "How do we develop an effective Human Robot Interaction model?"

    Another example, while I think this could partially be left up to the developers, MSRDS should provide some basic functions, again technologies that already exist in relatively robust form, like natural speech recognition and text to speech (I believe the latter may already be in there).  Again these technologies exist so why should developers have to spend time writing or integrating them again?

    Developers can't afford to spend their whole development cycle designing wheels and engines when they should be assembling the car.

    I'll be briefer in responding to your comments on the Experience and Business side (these sound like points from a PPT).

    In terms of Experience, I think it would be nice to go down a similar path to that of the smartphone appstore ecosystem.

    And on the Business side, you mentioned hardware and commoditizing component costs.  I have a lot of experience in the supply chains of the electronics industry and understand that while robotics won't drive component volumes and cost reductions in the immediate term, it shouldn't be much of an issue as standardized components from laptops & smartphones be it sensors or processors should be able to make up the bulk of what is needed.  You could probably also find motors and such from other volume products like power tools.  We've seen with Neato's robot LIDAR in the $10's while the cheapest LIDAR before that was at least in the $100's.

    In any case, I hope you find these comments useful.

    -NGM






    Thursday, January 27, 2011 5:28 PM
  • I have a little more fuel to add to this discussion.  Here's a link from the ROS.org project demonstrating the use of Microsoft's Kinect on a mobile robot.  I may sound like a broken record but there is a lot of activity on robotics technologies or the pieces of a practical puzzle, but without an environment to bring synergy to it all, we'll continue to tread water.

    Friday, January 28, 2011 8:41 PM
  • I just wanted to add that I've been working a lot around MSRDS for my PhD Thesis in coordination of multiple ground robots. 

    In that way, I have to say that the main drawback of using the RDS is the lack of services such as the already provided i.e. under ROS.org. 

    I hope to provide as much important services as I can. I will be uploading them at erobots.codeplex.com. I've already started to upload simple services I'm using, nevertheless something awful happened to my PC so that I haven't recovered to get on track again publishing more relevant and reusable services.

    Take a look at a very simple paper we developed for understanding the relevance of working with the system at:

     http://www.computer.org/portal/web/csdl/doi/10.1109/LARS.2010.18.

     

    Regards, 

    CC

    Tuesday, February 15, 2011 7:17 PM
  • I was going to write a long winded post on this topic, but it looked very whiney afterwards, so i will say only this. Every week i see a blog post about some cool new thing done in ROS, whilst the MRDS team haven't posted a news article in 7 months. Looking over the partner list on the website, that doesnt seem to be growing either. I am sure there are things going on behind the scenes, but this doesnt encourage new users to pick the MS solution, when both have similar learning curves (even if MS has a better UX).

    Tuesday, March 8, 2011 7:17 PM
  • Microsoft Research recently announced that they will release an SDK for Kinect this Spring:

    http://research.microsoft.com/en-us/news/features/kinectforwindowssdk-022111.aspx

    The Robotics Group has been evaluating a pre-release driver for Kinect with a view to releasing services for RDS. Obviously, we have to wait until the public release of the Kinect SDK before we can release anything because we have a dependency on the driver. Although this is not an official announcement of RDS support, I wanted to make it clear that we fully understand the importance of the Kinect as a sensor and the potential impact that it will have on robotics in general.

    Thanks for the feedback and your continued support.

    Trevor

     

    Sunday, March 13, 2011 12:40 AM
  • as I say in

    http://robotics.mech.upatras.gr/forum/viewtopic.php?f=10&t=19#p522

    (I guess Stathis won't have an issue reading the Greek text on that specific post [it's a mixed Greek-English forum])

    there's already a Kinect service for Robotics Studio shown in the following video by the SimplySim guys (who have done great work with Robotics Studio and have also collaborated with MS on the Sumo challenge at ImagineCup if I remember well):

    http://www.simplysim.net/blog/2011/01/11/simplycube-kinect-endless-possibilites/


    Microsoft MVP J# 2004-2010, Borland Spirit of Delphi 2001
    Thursday, May 19, 2011 12:43 PM
  • Keep your eye on the Kinect for Windows SDK from MSR :-)

    Trevor

     

    Saturday, May 21, 2011 6:23 AM
  • Thanks for the reminder on the name, looked it up and came accross

    http://www.winrumors.com/kinect-for-windows-sdk-beta-sign-up-now-live/

    which points to

    http://research.microsoft.com/en-us/um/redmond/projects/kinectsdk/ (BETA COMING SOON IT SAYS, hope so since end of Spring is very near ;-)

    (related feed is at http://research.microsoft.com/en-us/um/redmond/projects/kinectsdk/kdkrss.xml)


    Microsoft MVP J# 2004-2010, Borland Spirit of Delphi 2001
    Tuesday, May 24, 2011 8:59 AM
  • Interesting tweet: http://twitter.com/#!/KinectSDKTeam

    Hope this is the real twitter account of the Kinect SDK Team


    Microsoft MVP J# 2004-2010, Borland Spirit of Delphi 2001
    Wednesday, June 15, 2011 11:51 AM
  • Yes it does look interesting... ;-)

    Trevor

     

    Thursday, June 16, 2011 6:06 AM
  • This is interesting indeed http://research.microsoft.com/en-us/um/redmond/projects/kinectsdk/default.aspx Hope the robotics team is good to go with a CTP release soon :)
    Thursday, June 16, 2011 6:15 PM
  • Believe me, we have been waiting for this release just as anxiously as you have! However, we can't just ship an update immediately because we have to go through a whole set of testing and release procedures. These could not start until the Kinect for Windows SDK Beta was officially released. There are people working on it right now (including me). Hang in there, not long to go.

    Trevor

     

    Thursday, June 16, 2011 7:55 PM
  • you could always ship a Beta too ;-)
    Microsoft MVP J# 2004-2010, Borland Spirit of Delphi 2001
    Thursday, June 16, 2011 10:04 PM
  • Don't believe everything you see in a video. The Kinect uses structured light, not time of flight as the presenter says.

    Trevor

     

    Friday, June 17, 2011 12:51 AM
  • hehe Thanks Trevor, actually the best from that link is the download, which took me to download the beta release for kinect.

    http://research.microsoft.com/en-us/um/redmond/projects/kinectsdk/

     

    =D

    Friday, June 17, 2011 1:21 AM