OS startup

[babylon2233]

Startup IT company born everyday but it is very rare to see any of them starting their business with OS development. Maybe any of you want to share your view regarding this. At least there is one company I know from this forum, DiNS LLC . I see some potential in embedded market but it is almost impossible to gain anything from desktop OS market.

[AndrewAPrice]

A lot of companies were OS startups: Symbian, QNX, Microsoft (30 years ago), Red Hat.

[01000101]

At least there is one company I know from this forum, DiNS LLC .

I like this post already .

Startup IT company born everyday but it is very rare to see any of them starting their business with OS development.

This is probably due to several reasons,
1) OS development is significantly harder than application/website development when it comes to achieving a relatively similar result.
2) Instead of hiring alot of OK/IFFY developers for not-so-great pay and a few good ones to help crank out a software project, now companies need to hire the GOOD/GREAT developers for decent pay, and they are harder to find in the first place.
3) The competition is fierce: MS, Apple, Linux, CISCO, IBM, HP, which are all well established corporations.
4) Sometimes, it is just easier and more cost effective to just code whatever needs to be done, in a higher-level language with pre-made libraries (ex: games).

I see some potential in embedded market but it is almost impossible to gain anything from desktop OS market.

I agree. It would take a truly innovative design for a desktop OS to get enough support/funding to penetrate the market in even the slightest. I wouldn't dismiss the idea, but there have been many attempts, some went further than others, but almost all of them have never seen a store shelf. Embedded devices are great in the fact that they are (usually) specialized for only one or a few tasks, yet perform them very well. They also usually have functionality that is not found in desktop OS's, and therefore can penetrate the market fairly easily as they are unique and in a smaller marketing pool. DiNS fits in this genre as it will (one day) become an embedded device, and is pitched that way. Getting support was not all that difficult once I actually got a few sysadmins and lawyers to sit down and listen, but getting businessmen to actually discuss business with a (relative) kid is next to impossible. Reverting back to my previous statement, DiNS is a specialized device, it is part of an almost non-existent market pool and can therefore rise quicker and with very little resistance. I believe that innovation and unique ideas will lead more developers along the embedded/OS route as it offers almost unlimited flexibility and unlocks a great deal of computational power if designed right.

[babylon2233]

A lot of companies were OS startups: Symbian, QNX, Microsoft (30 years ago), Red Hat.

But Microsoft didn't start with developing operating system while Red Hat is not developing their own OS. Symbian and QNX are good examples anyway.

01000101, Do you have website for your OS?

[stephenj]

Generally speaking, programmers frequently face NIH syndrome. The typical question to ask is how much work would be saved by ad[ao]pting existing software that is available under acceptable terms?

In any event, if one can make money doing something they enjoy, more power to 'em.

Paul Graham is probably my favourite author on the subject of tech startup advice.

[prkeshri]

Ya sure It may be.......
Hey,I too wanna build an OS.......M a new comer..
Can You Help me......
From where do I start?
Languages I know or love to know:C++(a bit)
n Assembly language!!!

[01000101]

01000101, Do you have website for your OS?

nope, but I haven't really been looking into it either. I personally don't care if I have an official website while I'm in the development stages, maybe once I have a working product I will set one up. I don't have the time to code a good site while coding my OS.

[Dex]

I think you need to look for the next big thing, desktop OS have had there day, but if you find the next big thing, you are in with a chance.
As a example theres still no standard for a robot OS, Honda thinks it will be sell more robots than cars in 10-15 years.
And even M$ thinks so http://www.technologyreview.com/Biztech/17016/?a=f
http://msdn.microsoft.com/en-us/robotics/default.aspx
And with all the military spending on robots and unmanned vehicles etc, it maybe a good time for a startup.
This is the field i am working in.

PS: M$ First robot is a vacuum cleaner, early reports are that its the first M$ product, not to suck .

[Telgin]

I think you need to look for the next big thing, desktop OS have had there day, but if you find the next big thing, you are in with a chance.
As a example theres still no standard for a robot OS, Honda thinks it will be sell more robots than cars in 10-15 years.
And even M$ thinks so http://www.technologyreview.com/Biztech/17016/?a=f
http://msdn.microsoft.com/en-us/robotics/default.aspx
And with all the military spending on robots and unmanned vehicles etc, it maybe a good time for a startup.
This is the field i am working in.

PS: M$ First robot is a vacuum cleaner, early reports are that its the first M$ product, not to suck .

I'd imagine that there's no standard for a robot OS because there's no standard for robots! Robots vary tremendously after all, in both function and hardware. On the other hand it might be worth pursuing an attempt to make an extensible, modular OS designed to work on a number of embedded processor systems, and market it with the idea that future robot development might build off of it in a more standard way.

What the benefit of that would be though, I'm not so sure. Robots are so specialized most of the time that I doubt much could be shared and thus the OS wouldn't be able to provide much help in a general sense. Perhaps it could help by coordinating multiple processors, provide multitasking support, memory management, and a runtime environment, but hardware abstraction is going to be very difficult.

[babylon2233]

Honda's Asimo use VxWorks.

[babylon2233]

I think you need to look for the next big thing, desktop OS have had there day, but if you find the next big thing, you are in with a chance.
As a example theres still no standard for a robot OS, Honda thinks it will be sell more robots than cars in 10-15 years.
And even M$ thinks so http://www.technologyreview.com/Biztech/17016/?a=f
http://msdn.microsoft.com/en-us/robotics/default.aspx
And with all the military spending on robots and unmanned vehicles etc, it maybe a good time for a startup.
This is the field i am working in.

PS: M$ First robot is a vacuum cleaner, early reports are that its the first M$ product, not to suck .

I'd imagine that there's no standard for a robot OS because there's no standard for robots! Robots vary tremendously after all, in both function and hardware. On the other hand it might be worth pursuing an attempt to make an extensible, modular OS designed to work on a number of embedded processor systems, and market it with the idea that future robot development might build off of it in a more standard way.

What the benefit of that would be though, I'm not so sure. Robots are so specialized most of the time that I doubt much could be shared and thus the OS wouldn't be able to provide much help in a general sense. Perhaps it could help by coordinating multiple processors, provide multitasking support, memory management, and a runtime environment, but hardware abstraction is going to be very difficult.

I think you're wrong regarding specialize things as most embedded OS in the market today is modular and customizable.
What we need for a robot is a very basic OS that can be extended. So, things like A.I function will be implement on top of that basic OS.

[Telgin]

You probably have a point, but there is still a fair amount of distinction between an embedded device like a cell phone and a robot.

Of course, I suppose that the term robot needs to be defined in this case. If we're talking anthropomorphic robots intended to eventually mimic human behavior with complex AI and special complicated processing requirements (i.e. kinematics solvers for joints to move it approrpiately, 3D spatial analysis of environments to understand their shape, and so on), then I don't see how any generic OS can really help all that much. Like I said, it could provide basic services, but a specialized kernel made specifically for the purpose of assisting the robot would probably work better.

Simpler robots might not need so much. Ones that assemble cars for instance could probably use such a generic system without much difficulty if there were a standard.

Still, the main point of an OS is to provide hardware abstraction, processor and memory management, user interaction, and things of that nature. Like I said before, complicated and massively varied robots probably wouldn't make hardware abstraction all that worth the effort, and other OS support features could probably be built on a case by case basis.

But, I don't really know I guess. It's not like I'm in the industry or anything.

[prkeshri]

Is anyone going to help me out hw to start an Os-development???

[AJ]

Yes - http://wiki.osdev.org .

Please start a new topic if you are changing the subject and start by doing background reading (including forum rules).

Cheers,
Adam

[Dex]

A OS for a desktop OS is like you say to provide hardware abstraction, but i think this is the big advantage a robot startup would have, because of all the hardware drivers you need for a desktop OS, you have no chance.
But with robots, its differant, if you written a self learning AI, it could become the brains of the robot (so to speak), this does no need high graphics or great sound interface.
It could be made to be portable very easy, remember do not think interms of desktop OS, but rethink a robot OS.
The OS will be the self learning brains, that takes input and reacts to it, in a logical way.