Even for corporate usage it could be a benefit. Imagine how using BT to distribute patches and service packs could reduce the load on the MS update servers.
But those old planes are a long way from original spec. They'll typically run fairly recent avionics, so it's basically a new plane in an old airframe.
Anyway, cool things happened because back then, they had a plan, did their best, accepted the risks, and improved things as technology allowed. Now we're trying to get it 100% the first time. Why? Lawsuits, I imagine.
I think the WINE idea is a bad one. Apple would be much better served by software houses making native versions. Apple is all about the user experience, and I think they'll bend over backwards (or forwards) to help Adobe and friends port to OS Xi. See OS/2 for other reasons why WINE is a bad idea.
I can also see MS getting pissy about this move, so Xi versions of Office apps may come very slowly. I'd expect to see Apple putting some work into a native version of Open Office, possibly rebranded, or working with Sun on the Star Office brand.
I'd wager that Apple and AMD will talk though. This might inspire AMD to optimise gcc more for the the 64bit stuff to compete with Intel's compilers. Hypertransport and Opteron would make a lot of sense on a G5-level Apple workstation. Maybe Apple could talk to NVidia about chipset design too, with either Intel or AMD CPUs.
While basic consumer systems will likely be commodity hardware in a nicely designed package, anyone thinking of OS X on a whitebox workstation should remember both Sun and SGI have (or had) x86 products. I hope Apple shoots high with their professional products.
Just so you know, there's about a zillion free-for-home-use Linux antivirus applications. F-prot and Panda are just two of them. Personally, I run NOD32 on both Windows and Linux, but it aint free. It's good to see McAfee with a no-setup-drama on-access scanner though.
That highlights something I have trouble getting my head around, whether or not vision is used.
Even we (as drivers) don't know exactly where something is. The edge of a track in the desert is a good example - you take your best guess based on ruts or defoliage, and feel the response from the car when you get off the ruts. You make a correction to keep the ride as "smooth" as you can... still no good? You slow down, or try to find another route.
You *don't* need to know every rock on the road, just those big enough to cause problems. Another post talked about reducing the domain... that's what we do everytime we drive. We learn what the car can do, and ignore anything below a certain threshold. Check it out at your supermarket judder-bars - boys in low cars slow down, moms in suvs don't. The boys are processing more because their cars need them to. The moms aren't because their suvs soak it up for them. Neither of them calculates based on vehicle approach angle, they just learned what height and grade they can ignore. Better chassis + good domain reduction = easier task.
I was thinking that too. I'd even go as far as splitting the code up, and doing test-driven development.
* Define a data model for the IO and AI.
* Utilise an existing VR engine, and get a team working on the route navigation stuff. It doesn't have to be "real world" as the software doesn't know any different. Use topo maps or whatever, and work it until your virtual robot can navigate from A to B within the capabilities of the car and the world you give it, on the assumption that the IO system can produce the data model.
* Go find some test tracks, and work out the vehicle control details with a human in control. Ensure that the IO is what's needed for the AI to manage the car.
* Procure or produce the IO systems. Drive-by-wire is off-the-shelf. Most of the imagining is too, with some home-brew image processing on the top.
* Let a human plot the general course, then teach the car how to drive. Start with A to B stuff (gas+steering+brakes). Throw in a bog or a rock-field that needs to be analyzed further, add a hill that needs engine braking on the descent, etc. Make new tests that it can't do, then make it pass them.
* Rinse and repeat until happy.
* Profit!?
How about an OSDN Entry next year?;-) Of course, the OSS development process might give some clues to the competition...
For the visual sensors, because sensors don't rotate or change focus like eyes so different angles are needed, different viewing distances, etc.
Ah OK. I imagined focusable, rotatable cameras, on an isolated platform. I'm sure smarter folks than me have considered this, but I figured it would make sense to model terrain that way. Because that's how I do it. Maybe it's too slow.
For the visual sensors, because sensors don't rotate or change focus like eyes so different angles are needed, different viewing distances, etc.
Yeah, that's pretty much what I figured. Much the same as a person would do it. I was more getting at "optical ground recognition" with that question. Thinking about how people recognise a bog or ice or water without actually standing in it ("ooh, that ground's shiny/dark/textured/etc"). I guess you have sensors that can measure ground density or something.
You seems to have alot of faith in neural nets. They aren't the magical end all be all of AI you seem to think.
Not at all. I'm just using it as an example, although much of offroad driving is little more than function(scenario)=action. Of course there's higher level planning needed, for course mapping and so on, but I think that most of the details *can* be reduced to a functional level.
Take your Mom out in a truck, and tell her to drive through a small bog. She'll probably drive in, and get stuck. Now tell her about the diff-locks. She'll either get through, or drive in, get further, then get completely stuck. So you say, "Hey, it's 4wd, but still... how about finding an alternative?" So she drives around the bog. Now she can weigh up the "severity" (texture, size) of the bog against an approach to it - drive through locked up, or drive around. NN or Fuzzy Logic just seemed like relatively easy ways to learn substract rules.
I read some teams are using 7 or 8 fast CPUs here, and I wondered if maybe that was an indication of a) this is *really* hard, or b) too much information is being processed the wrong way.
There's a zillion ways to destroy a heavy vehicle, or parts thereof. Drivetrain seems to be a weak point in many offroad vehicles.
Your post just says to me, add some mics and some audio processing! When you drive, you listen to the car. You learn what's normal for a given speed or condition, and any waveform that's out of sync flags an error. You stop, look around, back up, and try something else. I had to explain this to friend of mine the other day - it had never occured to her to listen to a engine (or transmission) to hear the load.
If you really don't like audio, run a calc against axle speeds, engine RPM, clutch slip, figure out how much torque is where, and back off when something's close to breaking.
I'm a coder, not an AI or image-processing geek, so these might be dumb questions... but...
Why the need for so many sensors? I can understand a use for them in low-visibility, eg dust or darkness, but the current models seem excessive to a layman. I mean, can one not use steroscopic cameras (scanning the field, as our eyes do), run edge and shade detection over the frames, and generate 3D terrain models in real time?
How does a vehicle determin terrain density and route selection? Can terrain texture be estimated based on reflection or image matching, so the vehicle can decide not to drive over some water or a bog, for example?
Even a good human driver is going to get stuck in the deset without learning how to handle a truck offroad. Is it feasible to train a neural-net system to select a likely course, possibly with a set of hardwired rules as a base? Eg, make your own way, but don't sink the car.
I've no doubt this stuff is Hard, but much of this appears to be done via brute force...
I'll second that. Boomer is the hottie of the show. Starbuck is a little too butch (yeh, she's supposed to be) but they got it right with Boomer. Cute girl, good character, and well acted emotion. Cylon or not, I'd hit it.
Yeh, but Linux sales?! I'm sure Sun, SGI, IBM, HP et al Unices are all payed up, but how many Linux servers were bought and payed for? I'm having trouble thinking of any that weren't downloaded distros and either built from parts or converted Windows boxes.
I had a play with BrookGPU a while back, running parallel test jobs an AGP GF4 GPU, and PCI GFFX GPU. Worked well enough, but that was on PCI and AGP bus, and it killed the CPU trying to keep up with the GPUs. Probably needed bigger datasets to keep the GPUs busy or something...
Anyway, if there's an easy way to load-share accross these things using a single graphics context, I reckon it's got huge GPGPU potential.
Yes, piracy happens and probably costs them a significant amount of money but no amount of wrongdoing by a group of people justifies legislation or activity that infringes upon the rights of the innocent.
One small nit-pick. Piracy doesn't *cost* them anything other than what they spend to combat it.
Those terms annoy me almost as much as saying copyright infringment is theft.
I think it's more likely to push companies to rethink their legal strategies. I'd expect to see some trivial copy protection, and people getting sued (or at least threatened) for breaking it to upgrade their software. Even an EULA that specifically says, "Thou shalt not flash any model other than X" might be sufficient...
It's worse than that. Lets say you're a young professional out of school. This example is for New Zealand, but many others will be similar.
You earn a low-scale salary, and the govt takes 19.5% income tax (PAYE). Higher incomes attract higher taxation rates for those portions (33% and 39%). Also, you have a student loan, so they take an other 10% for that. So you get to take home 70.5% of your salary.
Then, everything you buy retail has 12.5% GST on it. So your dollars actually have only 87.5% of their value. You're getting 61.68% of your salary now. Many of your other purchases are taxed/levied additionally too. Petrol (Gas) being the most obvious one, and anything that's been imported has probably been taxed at least once as well.
Lets say you're saving some of what's left. Well, any interest you earn gets taxed at your income rate of 19.5%. Maybe you take another job to make ends meet? Well, that's got an additional 10% PAYE tax there, so you're down to 52.93% now, assuming you didn't jump into a higher threshhold...
A lawyer is sitting in a bar having a drink when a beautiful women sits down next to him. The lawyer seeing oppurtunity buys the women a beer and proceeds to hit on her. He then asks her, "Would you sleep with me for a million dollars?"
The women looks at him and says, "You know for a million dollars, sure."
The Lawyer then asks, "Would you sleep with me for 20 dollars?"
The women is instantly upset and yells, "Twenty dollars, what do you think I am some kind of whore?!"
The lawyer then looks at her and says, "Well, we have already established that fact. Now we are just negotiating."
Slashdot Mirror
So does this mean MS has a plan to buy out all Antivirus vendors who support Linux? Who's next?
Even for corporate usage it could be a benefit. Imagine how using BT to distribute patches and service packs could reduce the load on the MS update servers.
But those old planes are a long way from original spec. They'll typically run fairly recent avionics, so it's basically a new plane in an old airframe.
Anyway, cool things happened because back then, they had a plan, did their best, accepted the risks, and improved things as technology allowed. Now we're trying to get it 100% the first time. Why? Lawsuits, I imagine.
I was *sure* she was innocent, until I pondered one thing. How can you not notice your bodyboard bag has suddenly gained 4kg of weight?
I'm pretty sure even Xtra is outsourcing support to India. Either that or they've got a lot of immigrants working for them.
Being the best *anything* in the world is a huge chick magnet. Moreso is the confidence that comes with it.
I think the WINE idea is a bad one. Apple would be much better served by software houses making native versions. Apple is all about the user experience, and I think they'll bend over backwards (or forwards) to help Adobe and friends port to OS Xi. See OS/2 for other reasons why WINE is a bad idea.
I can also see MS getting pissy about this move, so Xi versions of Office apps may come very slowly. I'd expect to see Apple putting some work into a native version of Open Office, possibly rebranded, or working with Sun on the Star Office brand.
I'd wager that Apple and AMD will talk though. This might inspire AMD to optimise gcc more for the the 64bit stuff to compete with Intel's compilers. Hypertransport and Opteron would make a lot of sense on a G5-level Apple workstation. Maybe Apple could talk to NVidia about chipset design too, with either Intel or AMD CPUs.
While basic consumer systems will likely be commodity hardware in a nicely designed package, anyone thinking of OS X on a whitebox workstation should remember both Sun and SGI have (or had) x86 products. I hope Apple shoots high with their professional products.
Just so you know, there's about a zillion free-for-home-use Linux antivirus applications. F-prot and Panda are just two of them. Personally, I run NOD32 on both Windows and Linux, but it aint free. It's good to see McAfee with a no-setup-drama on-access scanner though.
In the future there will be homosexual robots
Not that there's anything wrong with that.
That highlights something I have trouble getting my head around, whether or not vision is used.
Even we (as drivers) don't know exactly where something is. The edge of a track in the desert is a good example - you take your best guess based on ruts or defoliage, and feel the response from the car when you get off the ruts. You make a correction to keep the ride as "smooth" as you can... still no good? You slow down, or try to find another route.
You *don't* need to know every rock on the road, just those big enough to cause problems. Another post talked about reducing the domain... that's what we do everytime we drive. We learn what the car can do, and ignore anything below a certain threshold. Check it out at your supermarket judder-bars - boys in low cars slow down, moms in suvs don't. The boys are processing more because their cars need them to. The moms aren't because their suvs soak it up for them. Neither of them calculates based on vehicle approach angle, they just learned what height and grade they can ignore. Better chassis + good domain reduction = easier task.
Badger badger badger.
I was thinking that too. I'd even go as far as splitting the code up, and doing test-driven development.
;-) Of course, the OSS development process might give some clues to the competition...
* Define a data model for the IO and AI.
* Utilise an existing VR engine, and get a team working on the route navigation stuff. It doesn't have to be "real world" as the software doesn't know any different. Use topo maps or whatever, and work it until your virtual robot can navigate from A to B within the capabilities of the car and the world you give it, on the assumption that the IO system can produce the data model.
* Go find some test tracks, and work out the vehicle control details with a human in control. Ensure that the IO is what's needed for the AI to manage the car.
* Procure or produce the IO systems. Drive-by-wire is off-the-shelf. Most of the imagining is too, with some home-brew image processing on the top.
* Let a human plot the general course, then teach the car how to drive. Start with A to B stuff (gas+steering+brakes). Throw in a bog or a rock-field that needs to be analyzed further, add a hill that needs engine braking on the descent, etc. Make new tests that it can't do, then make it pass them.
* Rinse and repeat until happy.
* Profit!?
How about an OSDN Entry next year?
For the visual sensors, because sensors don't rotate or change focus like eyes so different angles are needed, different viewing distances, etc.
Ah OK. I imagined focusable, rotatable cameras, on an isolated platform. I'm sure smarter folks than me have considered this, but I figured it would make sense to model terrain that way. Because that's how I do it. Maybe it's too slow.
For the visual sensors, because sensors don't rotate or change focus like eyes so different angles are needed, different viewing distances, etc.
Yeah, that's pretty much what I figured. Much the same as a person would do it. I was more getting at "optical ground recognition" with that question. Thinking about how people recognise a bog or ice or water without actually standing in it ("ooh, that ground's shiny/dark/textured/etc"). I guess you have sensors that can measure ground density or something.
You seems to have alot of faith in neural nets. They aren't the magical end all be all of AI you seem to think.
Not at all. I'm just using it as an example, although much of offroad driving is little more than function(scenario)=action. Of course there's higher level planning needed, for course mapping and so on, but I think that most of the details *can* be reduced to a functional level.
Take your Mom out in a truck, and tell her to drive through a small bog. She'll probably drive in, and get stuck. Now tell her about the diff-locks. She'll either get through, or drive in, get further, then get completely stuck. So you say, "Hey, it's 4wd, but still... how about finding an alternative?" So she drives around the bog. Now she can weigh up the "severity" (texture, size) of the bog against an approach to it - drive through locked up, or drive around. NN or Fuzzy Logic just seemed like relatively easy ways to learn substract rules.
I read some teams are using 7 or 8 fast CPUs here, and I wondered if maybe that was an indication of a) this is *really* hard, or b) too much information is being processed the wrong way.
Yeh, I'd have picked a BSD kernel too. OpenBSD makes political sense, in so far it's probably the most secure in the eyes of the geek community.
There's a zillion ways to destroy a heavy vehicle, or parts thereof. Drivetrain seems to be a weak point in many offroad vehicles.
Your post just says to me, add some mics and some audio processing! When you drive, you listen to the car. You learn what's normal for a given speed or condition, and any waveform that's out of sync flags an error. You stop, look around, back up, and try something else. I had to explain this to friend of mine the other day - it had never occured to her to listen to a engine (or transmission) to hear the load.
If you really don't like audio, run a calc against axle speeds, engine RPM, clutch slip, figure out how much torque is where, and back off when something's close to breaking.
I'm a coder, not an AI or image-processing geek, so these might be dumb questions... but...
Why the need for so many sensors? I can understand a use for them in low-visibility, eg dust or darkness, but the current models seem excessive to a layman. I mean, can one not use steroscopic cameras (scanning the field, as our eyes do), run edge and shade detection over the frames, and generate 3D terrain models in real time?
How does a vehicle determin terrain density and route selection? Can terrain texture be estimated based on reflection or image matching, so the vehicle can decide not to drive over some water or a bog, for example?
Even a good human driver is going to get stuck in the deset without learning how to handle a truck offroad. Is it feasible to train a neural-net system to select a likely course, possibly with a set of hardwired rules as a base? Eg, make your own way, but don't sink the car.
I've no doubt this stuff is Hard, but much of this appears to be done via brute force...
I'll second that. Boomer is the hottie of the show. Starbuck is a little too butch (yeh, she's supposed to be) but they got it right with Boomer. Cute girl, good character, and well acted emotion. Cylon or not, I'd hit it.
I don't argue that. I'm wondering how many more are in use that were never "sold" per se.
Yeh, but Linux sales?! I'm sure Sun, SGI, IBM, HP et al Unices are all payed up, but how many Linux servers were bought and payed for? I'm having trouble thinking of any that weren't downloaded distros and either built from parts or converted Windows boxes.
There are libraries for some of that here... http://graphics.stanford.edu/projects/brookgpu/ and here... http://libsh.org/
I had a play with BrookGPU a while back, running parallel test jobs an AGP GF4 GPU, and PCI GFFX GPU. Worked well enough, but that was on PCI and AGP bus, and it killed the CPU trying to keep up with the GPUs. Probably needed bigger datasets to keep the GPUs busy or something...
Anyway, if there's an easy way to load-share accross these things using a single graphics context, I reckon it's got huge GPGPU potential.
Yes, piracy happens and probably costs them a significant amount of money but no amount of wrongdoing by a group of people justifies legislation or activity that infringes upon the rights of the innocent.
One small nit-pick. Piracy doesn't *cost* them anything other than what they spend to combat it.
Those terms annoy me almost as much as saying copyright infringment is theft.
Also, check the pricing. A box of Win2K Pro costs *more* than WinXP Pro. They sure don't want folks runnnig the old stuff...
I think it's more likely to push companies to rethink their legal strategies. I'd expect to see some trivial copy protection, and people getting sued (or at least threatened) for breaking it to upgrade their software. Even an EULA that specifically says, "Thou shalt not flash any model other than X" might be sufficient...
It's worse than that. Lets say you're a young professional out of school. This example is for New Zealand, but many others will be similar.
You earn a low-scale salary, and the govt takes 19.5% income tax (PAYE). Higher incomes attract higher taxation rates for those portions (33% and 39%). Also, you have a student loan, so they take an other 10% for that. So you get to take home 70.5% of your salary.
Then, everything you buy retail has 12.5% GST on it. So your dollars actually have only 87.5% of their value. You're getting 61.68% of your salary now. Many of your other purchases are taxed/levied additionally too. Petrol (Gas) being the most obvious one, and anything that's been imported has probably been taxed at least once as well.
Lets say you're saving some of what's left. Well, any interest you earn gets taxed at your income rate of 19.5%. Maybe you take another job to make ends meet? Well, that's got an additional 10% PAYE tax there, so you're down to 52.93% now, assuming you didn't jump into a higher threshhold...
Bastards.
That reminds me of a joke...
A lawyer is sitting in a bar having a drink when a beautiful women sits down next to him. The lawyer seeing oppurtunity buys the women a beer and proceeds to hit on her. He then asks her, "Would you sleep with me for a million dollars?"
The women looks at him and says, "You know for a million dollars, sure."
The Lawyer then asks, "Would you sleep with me for 20 dollars?"
The women is instantly upset and yells, "Twenty dollars, what do you think I am some kind of whore?!"
The lawyer then looks at her and says, "Well, we have already established that fact. Now we are just negotiating."