Microsoft robotics closing its doors!
bomber
Posts: 297
RIP to EDDIE and now RDS! To add insult to injury they show a 3D model of the EDDIE platform with the screen showing the "Windows is shutting down..." screen.
http://spectrum.ieee.org/automaton/robotics/robotics-software/microsoft-shuts-down-its-robotics-group
http://spectrum.ieee.org/automaton/robotics/robotics-software/microsoft-shuts-down-its-robotics-group
Comments
Had they instead aimed their "robotics" research to anything linking computing with mechanics, they'd be making money already. Instead, they developed a set of software wrappers that have been done already. They never "owned" any robotics cred. They relied on third parties to create the hardware, and do all the robotics innovation. Problem is, anyone on Kickstarter can create a robotics OS. In fact, many have.
* For example, MS could have long ago developed kernels, interfaces, and hardware specifications for 3D machines. Those are robots.
* They could have, long ago, researched putting a Windows OS on board a quadcopter, controlling it, even taking pictures. Those are robots, too. What's more, the FAA is starting to relax their rules regarding commercial use of "drones." MS could have worked with a vendor -- possibly Parallax -- to create a flexible and affordable quadcopter robot that could eventually be leveraged in the commercial real estate, aerial photography, civil engineering, and other fields.
* They could have, long ago, used their leadership in motion capture optics to combine an industrial version of the Kinect with electro-mechanical assistance devices for physical therapy, sports training, mobility, and dozens of other practical and profitable business sectors. All of these are "robots" to one degree or another. They're simply not 100% autonomous. Complete autonomy is only one aspect of a certain kind of robotics.
Instead, Microsoft saw robots as having two wheels, wandering around the room exploring, and maybe doing a trick or two. Okay, so I'm being overly critical here of all of their research, but it pretty much boils down to this same many make when they think "robot." They're too stuck in creating the first real Robby or C-3PO. There are many smaller innovations to be made along the way.
We warned against building on MicroSoft's RDS years ago.
Sleep with MS and it will end in tears. Or at least severe pain in the nether regions.
Four hours and nine minutes... Feeling okay today?
Sorry, I was a bit slow off the mark there. It's been a long day. I'll try to do better in future.
Meanwhile, you seem to imply I am wrong. In which way might that be?
More seriously, what is going on with MS? Everyday I have been reading how MS has been firing a bunch of people and closing down this or that department.
I can not believe they are running out of cash just yet.
And, the cooperation with Microsoft was very productive as it furthered our evolution on these mid-size robot platforms. Specifically, having a high-volume customer allows us to invest in better tooling, molds, hardware improvements and often with a lower cost to the customer. Today, all of our customers get the benefits through the Arlo platform. ROS users seem to like Arlo as a favorite, too.
In sum, the work with Microsoft was good for Parallax and our customers.
Ken Gracey
I would also imagine they were doing some people a favor by shutting it down.
I saw Tandy Trower & troupe make an MSRS presentation at CES several years ago. Ironically the robot demo failed onstage, but I can believe that was due to trying to maintain a wireless connection in the noisiest e-noise environment imaginable. I got to chat with Tandy for several minutes after his demo and I showed him my venture bot. He was a very personable fellow. Hope he's doing good things at Hoaloha.
Microsoft's usual approach is to build the underlying tools and wait for others to develop working products with them. I don't think that works with hardware, and their past experiences have shown that. In addition to Eddie, there were several third-party platforms that had support, one way or another, in RDS. But I think many suffered from a very loose integration, and a roadmap that was hard for the lower school grades to understand and follow. I'm not sure many teachers saw the benefit of adding RDS as a layer on top of what they already had in the visual programmer that comes with LEGO NXT, for example.
I personally think MS should have targeted 5-6 growth markets, perhaps some of the ones I mentioned earlier, and seeded development of hardware through cooperative -- but full-scale -- investments and co-development agreements. I think focus into specific markets would have helped them.
So, the upshot is that despite the hard work and efforts by those who actually crafted the tools, the lack of a distinct commercial goal from the start has ended with yet another example of how robotics is a no where field. Which of course is far from true.
I think if you have a good sysadmin, it doesn't matter what OS you run
Any commercial success will be something more like Roomba. Not a vacuum necessarily, but an affordable, simple robot optimized to do a single job well and reliably.
MS could have tackled the robotic home care market with some Microsoft-branded hardware addons, like what Revolution tried with Northstar. Software-only doesn't cut it, and not everyone has the engineering to create the hardware for self-mapping robots. Their hardware could have provided IoT WiFi, some sort of depth vision and mapping, and perhaps a homing system. All of this in one box. Then others could buy or license that part, and add it to their platform, which would provide their unique offering in this market.
In the case of the now-burgeoning UAV market, MS might have worked with Parallax to create board-level firmware that worked with custom Windows-based smart phones and mini tablets that could, in turn, work on anyone's air frame. Their corporate clients would have bought the hardware from them, or licensed it for manufacture. I believe with its 8 cores the Propeller would have been the ideal non-ARM selection for the hardware controller. It's as real time as it gets.
Alas, all of this is conjecture of what might have been. I sigh whenever a once-vaulted robotics venture crashes and burns. Especially from someone as high profile as Microsoft. It makes it that much harder for the next guy with a brilliant idea to get funding.
I'm on the case, Gordon. My venture bot has become my top priority now. The object of my fixation. It's now or never. Wish I could say more.
-Phil
As far as MSRS - We got Arlo from this process - don't forget it!
Shouldn't robot platforms alternate male/female names, as do hurricanes?
It started with MadeUSA, then Eddie, which was fine. But then Arlo? How about Arlene instead?
Am I too late to the party with this observation? Why does this stuff bother me so?
If that's correct, we'd still have Arlo, but as Ken points out, the collaboration with MS increased some sales there, allowing for some plant expansion and ongoing development, perhaps in things like the encoders. And maybe helped a salary or two, which is never a bad thing.