Here’s how Microsoft Robotics Studio is getting used…

About a week ago we had an interesting presentation at Microsoft Research given by Rod Grupen from the Laboratory for Perceptual Robotics, Department of Computer Science, University of Massachusetts Amherst (UMass) – the lecture was very interesting, and Rod is an excellent speaker – Rod brought some students from the college and an interesting balancing robot called Ubot5 – The robot is running on an x86 based platform, which means it can run Windows XP/Embedded, and therefore could run the CCR/DSS technologies from Microsoft Robotics Studio – in fact, the students spent just two days getting the robot up and running (or at least balancing!) using Robotics Studio – here’s a video of the robot “doing it’s thing”, being controlled by the Robotics Studio dashboard.

Video: UBot5 – Self Balancing Robot using Microsoft Robotics Studio

Or, how about a self driving car that took part in the Darpa Urban Challenge? – The Princeton Autonomous Vehicle Engineering (PAVE) team are running MSRS across 5 dual-core systems, here’s how Robotics Studio is being used – and of course, video of the vehicle doing its thing (more video here).

– Mike

Comments (6)

  1. Here’s how Microsoft Robotics Studio is getting used… Something is going on here…how long before…

  2. W says:

    I wonder if that’s using the traditional closed loop control mechanism or the more advance state space.  I wonder how fast robotics studio is.  Could it meet the high speed timing requirements of a bipedal closed feedback robot.  

  3. Thanks Mike for the great post. We have a set of cool stuff including robots, user groups, partner releases, etc. using MSRS on our community page [1] and will get more from the UMass guys shortly.



  4. George Chrysanthakopoulos says:

    Regarding performance: MSRS, on a "standard" PC (2GHZ dual core) can do about 1.5 million messages per second between two independent CCR components (Basically two classes posting messages on ports, each on a different thread/core).

    Between DSS services, same node, with full cloning of each message, we can do about 100 thousand messages / sec. Across machines we can do about 6,000 msgs/sec.

    So its fast enough for LQR, PID, Kalman discrete time filters with very high update rates.

    We have partners that are using the CCR/DSS for the high cycle rate control logic.

    In this case however, the ubot5 has an FPGA that does the high cycle rate control inside (the balancer/inverted pendulum), and then MSRS is used for high level orchestration and higher order control. Also MSRS is used for the articulated arms, implementing the splines, etc.

    The ubot5 guys will post more details/demos soon as well. Btw they were following this two-tiered approach before they ported to MSRS, since we believe its the right path: isolate tight control at some different simpler hardware level, use MSRS and a general purpose OS, at a higher level.

    Hope this makes sense

  5. [3:14PM] I’m here now in the Faculty Summit Academic Track. Jennifer Perret has just given an overview

  6. wotudo says:

    It seems an age already but only last month I was working in Redmond for a week. During this time I got