In the early 1980s Doug Lenat used an evolutionary AI system to win a sci fi naval battle contest. His system, Eurisko, designed unexpected large fleets of ships that defeated more conventional systems designed by other players, overwhelming them with numbers.
http://aliciapatterson.org/sto...
So when the Navy uses AI to run the ships, AND to design them, might have something.
No one has mentioned so far the growing importance of academic conferences over journals. In some fields, particularly computer science, the problems with journal articles cited by the original poster have led to a greater emphasis of conferences.
Here's some assertions I'm not going to fact check now but have seen in the past few years:
a) In computer science, the major conferences have a higher rejection rate than journals.
b) Conference papers are submitted, accepted, and appear in under one year. Journals articles in CS take an average of 32 months to appear.
c) Traditionally the model is to publish results in a couple of conferences and then sum it up in a significant journal paper. A number of cs academics have gone to a conference first and only strategy, bypassing journals.
d) Conferences are a perfectly valid venue to cite in cs.
e) ACM and IEEE have put out position papers to defend the importance of conferences when campuses consider faculty for tenure (since faculty in other fields may not understand.)
f) Conferences often bypass the for profit publishers (not that IEEE is a charity).
g) In a flip move, many conferences proceedings now appear as special issues of a journal.
Combine all this with a move to electronic distribution (no printed conference proceedings, just a CD/DVD) moves the process to an online model.
So some of the problems cited by the original poster have been mitigated in part by a move to conferences as a significant academic outlet. And the conference itself, with the chance to meet and talk with the authors, builds community and confidence in results.
Computer scientest David Gelernter proposed a organization for all the material we produce as a "lifestream", a date-stamped list of all electronic documents in our lives. When I first heard of I thought, naa, too ill-organized.
But, I find my email working in exactly the way he proposed. My email package provides the best database I have of my work and communication. Searchable by date, correspondent, content, subject; control-click to organize by date, sender, header; automatic filters to sort by same; subfolders; attachments of all kinds accessable by the search; and I can add to it from anywhere by emailing myself. I use email to mainain to-do (email myself), I use email to maintain a calendar of past activities by searching for email on the topic (when did we do X?) , I use email to store minor documents and search for them as attachments. By using pop and downloading all email to my harddrive, I have no limitations of an account.
So, while dubious about "lifestreams", I've backed into it as the core of my work habits.
Here's what I remember about the perception of light. Not a professional in the area.
The color receptors (cones) are least sensitive to blue light. Ever see a blue firetruck?
The green receptor is most senstive (hence the new greenish firetrucks), the red second and the blue third.
The b&w low-light receptors (rods) are least sensitive to red light, hence the use of soft red illumination in WWII pilot briefings before night flights. If you're buying one of those small led flashlights to use at night when you want to still be able to see effectively in the semi-dark when the flashlight is off, buy a red version.
Those blue leds just must be putting out lotsa lumens.
Some stations require it, some don't - I don' t think Mobil aimed this restriction at cracked RFID cyphers, but rather at stolen speedpasses. Same way some stations in small midwest towns still don't required payment in advance due to a moderately honest local population.
Here's three free, current programming environments that are suitable for introducing programming (that no one seems to have mentioned yet.)
1. Design by Number
Created by John Maeda of MIT, this is a very simple graphics-oriented programming language. Maeda created it for artists and there an associated book. Like a sparse Logo, it keeps everything to a bare minimum. Has a web applet that allows interpreted programming to try it out.
DBN web site
2. Processing
DBN is no longer maintained, and a more complex graphics language emedded in Java (with a single-line interpreter for ease of use) has been developed by Ben Fry and Casey Reas of MIT.
Processing web site
3. Robocode
Developed at AlphaWork at IBM, this is a Java environment for programming your own virtual robots that then shoot each other. Has a 2d battle arena in which little tanks move and shoot. (Classic idea, just a nice implementation). You can program as much or as little intelligence as you wish. Designed for teaching Java.
Robocode web site
What is the environmental trade-off with digital frames vs. paper photos?
My father took 10,000 slides over the years, and between his slides and print photos taken by members of the family, I have many, many large boxes of pictures from pre-1900 to now. We've also kept many of the cameras, back to the Kodak boxes and the accordian fold-outs.
The resource costs of the photos and their processing has probably been large, and the photos and slides just sit in albums and boxes. Particularly the slides - it takes a projector and some time to see them properly.
I've gone to a digital system now and have thousands on my computer that require no additional resources (besides disk space) and can be scanned through quickly, dumped into slide shows, and otherwise used more easily. No piles of albums, no processing costs, no boxes, no slide projectors.
If we implement digital frames and a supporting environment correctly, so we can zip our digital archives to digital frames at work, at home, and at friends, then the resource cost might be less over a lifetime, even with the electricity and tech-production costs. What is the annual electrical use of a digital frame?
But, this will take attention to environmental costs and their proper calculation. Not sure what the balance is in this case, but it isn't obvious that the new system inherently uses more resources.
Dr. Rubin has already done this in a class. In
CS 600.643: Advanced Topics in Computer Security
he has student research methods for malicious code hiding and then as groups, try to hid code to rig the vote in a program. Once done, the groups switch programs and try to find the malicious sections. Groups were given one clean, one dirty and one version picked at random.
Results? Very few sections of malicious code were found, even in this highly controlled environment. The graduate students proved very adept at obsfucation.
I hope Dr. Rubin reports on this in the literature because the results sound fascinating from a keynote address he gave last Friday (Oct. 15 at CCSCE 04). Sounds like a great class assignment.
Here we agree. It isn't research vs. teaching - it's only research, or only teaching, vs. a good balance between the two. And I agree that universities getting patents, etc, from publically funded grants stinks. And faculty that focus only on research goes too far. And presidents and provosts that are too egotistical are unfortunate drains. We agree on all this.
what ever happened to recruiting good teachers who were interested in teaching instead of stroking their egos when it comes to their research?
All schools want to do this, but it's tough. Faculty that succeed at research usually have an ego - to be publishing articles they have to be convinced they know more than anyone else in their area. Look at E.O. Wilson's bio from Harvard - he thought that in his 30/40s that if he applied his work ethic, he could break marathon race records (he didn't.) On the other hand, faculty that focus heavily on teaching often lose the research and innovation edge.
This isn't to say the balance can't be struck - it's just difficult.
the fix is quite simple. schools that receive public monies shouldn't have such a focus on research. leave research to the private unis, and leave publicly funded schools to teaching. if a student or his parents don't like the quality of education at the cheaper public schools, then they can take out a loan and send johnny or alice to yale.
The major state universities - the land grant institutions - were established for the dual purpose of education and research. That's part of their establishing mission. They can't abandon it. And, as someone who has taught at big public and mid-sized private, the research component is essential to keeping faculty that teach first rate classes. And that invent things. And start new companies. And attract students that will create firms, like, oh, say, Google.
And don't forget that for every dollar in "soft money" there's usually 50 cents in overhead that goes into the campus coffers and decreases the tax that needs to come out of your pocket.
If you don't do research, your teaching gets stale and irrelevant. The best teachers are those who can handle both. Any idea that you can take research away from the big state Us and improve their teaching is misguided.
Ah, but parents are already doing that. The University oF Maryland and other state schools have seen the GPAs and SATs of incoming students increasing greatly as families figure out exactly your point - that a state education at a first rate university, particulary one that has had a number of inititatives to "make the big school small", has a better price performance ratio than semi-Ivy private schools.
Besides, many of the schools with fast growing tuition are the "State U"s.
it's pretty pathetic that the "academic tradition" of openness at universities (i'm speaking of public state schools, not private unis which can do whatever they want for all i care) is slowly being pushed to the wayside. universities are no longer run like places of education, but more like businesses. business has no place in education.
Funding at state universities is problematic.
This year the University of Maryland - College Park gets more money from research grants (> $300M) than it does from the state subsidy.
These schools can't keep education affordable for in state students if the states continue to cut their budgets. In fact, some schools are decreasing the difference between in and out state tuition partly on the logic that if the state is going to cut subsidies, then in state students should bear those cuts.
One joke is that state universities are no longer state supported, but state molested institutions. The state gives less money but continues to regulate the schools. In South Carolina the governor has offered state universities the choice to opt out - go private - by giving up their state subsidy and state oversight. The boundaries between state and private institutions is blurring.
Raising tuition and focusing on research (ie, by constructing research buildings), and behaving more as commercial entities, are understandable reactions to the finanical climate of state schools. In this case, it means charging students for Napster rather than trying to eat the costs.
So while you may be very critical of your school's president, the actions you comment on may be driven by strong financial needs rather than an abandonment of social priorities.
Computer science is a new discipline and career which has been on a long growth curve since its inception, but boom-bust cycles are typical of engineering disciplines and there is no reason CS should be an exception.
Older slashdot readers will remember the aerospace engineering bust of the 1970s ("will the last person leaving Seattle please turn out the lights"), and there was an EE bust then too - http://www.engtrends.com/InsideEE/Article06a/. Engineers in resource extraction (ie, oil) see boom-busts related to demand, as do those in construction.
For CS PhDs we've gone through a couple of minor boom-bust cycles already. Those graduating in the 1970s and early 80s had easy times finding a job - then, late 80s the market was tighter, fewer positions - then, late 90s, PhDs had lots of options so schools had a hard time recruiting - and in the past two years now it's becoming easier to recruit.
I'm tempted to think of these cycles like the predator-prey population cycles (you know, lots of bunnies, then lots of foxes, then fewer bunnies, then fewer foxes, then repeat). It's just part of the engineering field. The key is that if you love your engineering discipline, and are good at it, you will find a job. If the discipline becomes familiar with the cycle, then we can discourage weaker candidates during boom years and encourage strong candidates during the bust years. Schools can't buy too much into the current cycle.
Are there fundamental changes in the discipline that would make the boom-bust cycle different this time? Increasing consolidation of firms, more barrier to entry from patents, more CS/programmers trained overseas? Maybe. CS is, after all, all about automating tasks and if we get too good at it, we can impact a lot of jobs. But, remember, life is NP-complete - there will always be more to do.
One of our students does exactly this, apache and all.
Has the wireless card in, so he can control the pc audio from anywhere on campus. You get a touch screen control that roams. Freaks out his roommate when he sets the music before he arrives at the door.
There's lots of starter code for computer vision tasks for those who want to play with the project. A major web page with links to research groups and software sources is
http://www-2.cs.cmu.edu/~cil/v-source.html
If you want to use intel processors, then there's a library of routines for many vision tasks supported by Intel at:
I want to say I fully respect the moral objection to working on DARPA projects, and consider it valid.
But, when I was working on the first DARPA ALV (automous land vechicle) project in the 1980s, you really couldn't draw a clear line between DARPA research and research supported by other funding sources, like NSF and NIH.
One of the grad students wrote a parody of a tech report generating program that would create a new tech report given a few key words and the funding agencies. Input like "blob", "identification", "snake", "NIH" would generate a tech report on "Finding blob shape tumours using snakes" - change "NIH" to "DARPA" and the title would be "Finding blob shaped targets in side looking radar".
The overlap between ideas for miilitary applications, and non-military applications, was so extensive that you couldn't untangle them. One research student was determined not to accept military funding, and didn't, Stiill, I was in the tech report room the day the DARPA tech officer came to get a few reports, and he made a point of selecting reports by that graduate student. "Great work, gotta follow it" was his line.
So, while I fully accept and respect personal decisions not to accept military funding, any open work to advance autonomous robots will end up in weapons. That's a consequence of the research area, not the funding source. Those in this area have to enter it with no pretenses.
G.N. Hardy, in his great memoir, "Apologies of a Mathematican", proudly stated that he never worked on a military project even during the time before and during WWII. C.P. Snow, in his preface to that book, noted that while Hardy never worked on such a project, he trained all the (Cambridge?) mathematicans in the British service. Science is the ultimate open source endeavor.
Actually, Apple computer lost and had to pay $26 million, but I think as a result of that settlement they got out from under the original agreement.
In the original Apple-Apple settlement from the first suit, Apple computer agreed to stay out of music. When they added microphones and went that direction, Apple music sued again and won, but I seem to remember that Apple computer was then freer to invovate. If you care about the details you can start with
They are different things, but you're missing the relationship between the two.
When he invented the Turing machine, he included the concept of a universal Turing machine (UTM) that could simulate the behavior of any other Turing machine. You only really needed one UTM.
Since he invented the UTM, lots of people have invented other computing methods and devices (like, AMD chips). Over the years all those new methods have been proven to be no more powerful than Turing original UTM - ie, a UTM can emulate any Perl, C, Java, AMD machine code, Recursive function, or etc. Very, very slowly, yes, but a UTM gets the job done. See the Church-Turing thesis for background.
If you go back and read Turing's original 1936 article on the UTM (available online), you'll note the importance of the UTM being a state machine. Read his 1954 article on the Turing test (also online), and you'll find an echo of the logic applied to thinking about human brains.
So what Turing was asking can be viewed in part as whether a UTM could emulate the behavior of the human mind (the strong AI hypothesis).
These aren't "totally different things" - there is a strong relationship. Hey, read the articles, they're accessible with a little work.
No, redragon, you've missed the computer science content of the previous post. Turing invented a universal computing device (ie, the Turing machine) that could emulate any other. Not a universally intelligent machine - you've mangled the term.
The question is whether the wetware in our brain has some computational power not possessed by a Turing machine equivalent device. If we don't have "extra" computing features, then our brains can be emulated by a Turing equivalent machine. Roger Penrose and others have attacked AI by arguing that the brain has extra computational power.
In some sense, we are talking about a simulating a human with a computer.
The iLife CD with all three is $14.95 special intro education price, so I opted for the snailmail solution.
Have them installed when I get home? Like I have time to use them, anyway....
Bill Gates and company figured this out when Java was released.
This article is much more interesting that that simple observation, though.
Nice, but what about the Army's ALV project?
on
Autonomous Race Cars
·
· Score: 3, Interesting
In the 1980's President Reagan had the Star Wars project, which everyone has heard about, but there was also $600M in funding for other projects including the autonomous land vehicle (ALV). A number of schools worked on this, but the most successful was CMU with a number of robot vehicles that could go up to 60 mph and drive on uncontrolled roads - one drove cross country in a demo. ALVINN was one name used for this system. As a result of Sept. 11th there is renewed US research funding for this area. Autonomous flying vehicles, like the Predator in Afganistan, also came in part out of the 1980s funding from DARPA - we just haven't seen the similar successful application of the land version.
The key to this particular competition seems to be the size and weight limits on the cars. This is a contest for little RC-style cars. The technology exists to go much faster, but not in this form factor.
Using it to use overlapping images to increase the effective resolution of the camera. This is called
"super resolution".
Computer vision types have been doing this for years.
Shmuel Peleg of Hebrew University has done some good work and had the work show up in commerical products, including Videobrush - you could take a webcam, wave it around, and in real time get a mosaic. In 1995 (I think.) Don't know if you can still get that product.
Do a google search for him and you'll find his home page and superresolution papers (Peleg and Irani is
an accessible paper and one of the first - the concept they used, however, comes from Bruce Lucas' thesis at CMU.
Applications include: combining NASA satellite images of the Earth to get higher resolution, ditto for images of the human retina; and, a personal favorite, smoothing images of the system used at the Superbowl a couple of years ago where they had 75 cameras and could show any play from any angle in liive video. That was done by Takeo Kanade, Luca's' advisor.
I think you're right, Camera Link faces an uphill battle with the other standards (why use a new interface when you can do anything you want with Firewire?)
But, I did find these supporting companies which includes Sony. There was a recent Slashdot post on how new technologies can challenge existing standards, and one claim was if the technology can find a small niche market not claimed by the big companies, it can survive - and machine vision is such a specialized, niche market.
* 3M Interconnect Solutions Division
* ADIMEC Advanced Image Systems BV
* Alacron, Inc.
* ATMEL Grenoble
* Basler Vision Technologies
* BitFlow, Inc.
* Cognex Corporation
* Cohu, Inc., Electronics Division
* Coreco Imaging
* DALSA Corp
* Data Translation, Inc.
* Datacube, Inc.
* Duncan Technologies, Inc.
* Epix
* Euresys S.A. Corporate Headquarters
* i2S - Line Scan
* illunis LLC
* Intercon 1
* JAI A-S
* Leutrek Vision, Inc.
* Matrox Electronic Systems Ltd.
* National Instruments
* PerkinElmer Optoelectronics
* PULNiX America, Inc.
* Sony Electronics Inc.
* Toshiba America Information Systems/ISD
* TVI Vision
* Xybion Electronic Systems
There is an emerging standard intended to replace firewire for high quality, industrial and scientific digital video cameras (ie, high end web cams.)
The standard is called Camera Link and you can find a technical PDF on it with a quick google search.
Camera link was designed specifically for connecting digital cameras so, according to the stuff I've seen so far, there are line scan cameras in camera link but not in firewire, camera link cameras are higher bandwidth than firewire cameras (up to 2.38 GB), and camera link is supported by the industrial (as opposed to consumer and artistic) camera and frame grabber community
I don't know much more than this, so I'm not advocating Camera Link, just putting it in this discussion for those who haven't heard of it.
Slashdot Mirror
In the early 1980s Doug Lenat used an evolutionary AI system to win a sci fi naval battle contest. His system, Eurisko, designed unexpected large fleets of ships that defeated more conventional systems designed by other players, overwhelming them with numbers. http://aliciapatterson.org/sto... So when the Navy uses AI to run the ships, AND to design them, might have something.
No one has mentioned so far the growing importance of academic conferences over journals. In some fields, particularly computer science, the problems with journal articles cited by the original poster have led to a greater emphasis of conferences.
Here's some assertions I'm not going to fact check now but have seen in the past few years:
a) In computer science, the major conferences have a higher rejection rate than journals.
b) Conference papers are submitted, accepted, and appear in under one year. Journals articles in CS take an average of 32 months to appear.
c) Traditionally the model is to publish results in a couple of conferences and then sum it up in a significant journal paper. A number of cs academics have gone to a conference first and only strategy, bypassing journals.
d) Conferences are a perfectly valid venue to cite in cs.
e) ACM and IEEE have put out position papers to defend the importance of conferences when campuses consider faculty for tenure (since faculty in other fields may not understand.)
f) Conferences often bypass the for profit publishers (not that IEEE is a charity).
g) In a flip move, many conferences proceedings now appear as special issues of a journal.
Combine all this with a move to electronic distribution (no printed conference proceedings, just a CD/DVD) moves the process to an online model.
So some of the problems cited by the original poster have been mitigated in part by a move to conferences as a significant academic outlet. And the conference itself, with the chance to meet and talk with the authors, builds community and confidence in results.
But, I find my email working in exactly the way he proposed. My email package provides the best database I have of my work and communication. Searchable by date, correspondent, content, subject; control-click to organize by date, sender, header; automatic filters to sort by same; subfolders; attachments of all kinds accessable by the search; and I can add to it from anywhere by emailing myself. I use email to mainain to-do (email myself), I use email to maintain a calendar of past activities by searching for email on the topic (when did we do X?) , I use email to store minor documents and search for them as attachments. By using pop and downloading all email to my harddrive, I have no limitations of an account.
So, while dubious about "lifestreams", I've backed into it as the core of my work habits.
The color receptors (cones) are least sensitive to blue light. Ever see a blue firetruck?
The green receptor is most senstive (hence the new greenish firetrucks), the red second and the blue third.
The b&w low-light receptors (rods) are least sensitive to red light, hence the use of soft red illumination in WWII pilot briefings before night flights. If you're buying one of those small led flashlights to use at night when you want to still be able to see effectively in the semi-dark when the flashlight is off, buy a red version.
Those blue leds just must be putting out lotsa lumens.
Some stations require it, some don't - I don' t think Mobil aimed this restriction at cracked RFID cyphers, but rather at stolen speedpasses. Same way some stations in small midwest towns still don't required payment in advance due to a moderately honest local population.
Here's three free, current programming environments that are suitable for introducing programming (that no one seems to have mentioned yet.)
1. Design by Number
Created by John Maeda of MIT, this is a very simple graphics-oriented programming language. Maeda created it for artists and there an associated book. Like a sparse Logo, it keeps everything to a bare minimum. Has a web applet that allows interpreted programming to try it out.
DBN web site
2. Processing
DBN is no longer maintained, and a more complex graphics language emedded in Java (with a single-line interpreter for ease of use) has been developed by Ben Fry and Casey Reas of MIT.
Processing web site
3. Robocode
Developed at AlphaWork at IBM, this is a Java environment for programming your own virtual robots that then shoot each other. Has a 2d battle arena in which little tanks move and shoot. (Classic idea, just a nice implementation). You can program as much or as little intelligence as you wish. Designed for teaching Java.
Robocode web site
My father took 10,000 slides over the years, and between his slides and print photos taken by members of the family, I have many, many large boxes of pictures from pre-1900 to now. We've also kept many of the cameras, back to the Kodak boxes and the accordian fold-outs.
The resource costs of the photos and their processing has probably been large, and the photos and slides just sit in albums and boxes. Particularly the slides - it takes a projector and some time to see them properly.
I've gone to a digital system now and have thousands on my computer that require no additional resources (besides disk space) and can be scanned through quickly, dumped into slide shows, and otherwise used more easily. No piles of albums, no processing costs, no boxes, no slide projectors.
If we implement digital frames and a supporting environment correctly, so we can zip our digital archives to digital frames at work, at home, and at friends, then the resource cost might be less over a lifetime, even with the electricity and tech-production costs. What is the annual electrical use of a digital frame? But, this will take attention to environmental costs and their proper calculation. Not sure what the balance is in this case, but it isn't obvious that the new system inherently uses more resources.
Dr. Rubin has already done this in a class. In CS 600.643: Advanced Topics in Computer Security he has student research methods for malicious code hiding and then as groups, try to hid code to rig the vote in a program. Once done, the groups switch programs and try to find the malicious sections. Groups were given one clean, one dirty and one version picked at random.
Results? Very few sections of malicious code were found, even in this highly controlled environment. The graduate students proved very adept at obsfucation.
I hope Dr. Rubin reports on this in the literature because the results sound fascinating from a keynote address he gave last Friday (Oct. 15 at CCSCE 04). Sounds like a great class assignment.
This isn't to say the balance can't be struck - it's just difficult.
And don't forget that for every dollar in "soft money" there's usually 50 cents in overhead that goes into the campus coffers and decreases the tax that needs to come out of your pocket.
If you don't do research, your teaching gets stale and irrelevant. The best teachers are those who can handle both. Any idea that you can take research away from the big state Us and improve their teaching is misguided.
Besides, many of the schools with fast growing tuition are the "State U"s.
This year the University of Maryland - College Park gets more money from research grants (> $300M) than it does from the state subsidy.
These schools can't keep education affordable for in state students if the states continue to cut their budgets. In fact, some schools are decreasing the difference between in and out state tuition partly on the logic that if the state is going to cut subsidies, then in state students should bear those cuts.
One joke is that state universities are no longer state supported, but state molested institutions. The state gives less money but continues to regulate the schools. In South Carolina the governor has offered state universities the choice to opt out - go private - by giving up their state subsidy and state oversight. The boundaries between state and private institutions is blurring.
Raising tuition and focusing on research (ie, by constructing research buildings), and behaving more as commercial entities, are understandable reactions to the finanical climate of state schools. In this case, it means charging students for Napster rather than trying to eat the costs.
So while you may be very critical of your school's president, the actions you comment on may be driven by strong financial needs rather than an abandonment of social priorities.
Older slashdot readers will remember the aerospace engineering bust of the 1970s ("will the last person leaving Seattle please turn out the lights"), and there was an EE bust then too - http://www.engtrends.com/InsideEE/Article06a/. Engineers in resource extraction (ie, oil) see boom-busts related to demand, as do those in construction.
For CS PhDs we've gone through a couple of minor boom-bust cycles already. Those graduating in the 1970s and early 80s had easy times finding a job - then, late 80s the market was tighter, fewer positions - then, late 90s, PhDs had lots of options so schools had a hard time recruiting - and in the past two years now it's becoming easier to recruit.
I'm tempted to think of these cycles like the predator-prey population cycles (you know, lots of bunnies, then lots of foxes, then fewer bunnies, then fewer foxes, then repeat). It's just part of the engineering field. The key is that if you love your engineering discipline, and are good at it, you will find a job. If the discipline becomes familiar with the cycle, then we can discourage weaker candidates during boom years and encourage strong candidates during the bust years. Schools can't buy too much into the current cycle.
Are there fundamental changes in the discipline that would make the boom-bust cycle different this time? Increasing consolidation of firms, more barrier to entry from patents, more CS/programmers trained overseas? Maybe. CS is, after all, all about automating tasks and if we get too good at it, we can impact a lot of jobs. But, remember, life is NP-complete - there will always be more to do.
One of our students does exactly this, apache and all. Has the wireless card in, so he can control the pc audio from anywhere on campus. You get a touch screen control that roams. Freaks out his roommate when he sets the music before he arrives at the door.
http://www-2.cs.cmu.edu/~cil/v-source.html
If you want to use intel processors, then there's a library of routines for many vision tasks supported by Intel at:
http://www.intel.com/research/mrl/research/opencv/
Here's what's fun about opencv - it's an effort to make vision algorithms as accessible to programmers as opengl makes graphics algorithms.
But, when I was working on the first DARPA ALV (automous land vechicle) project in the 1980s, you really couldn't draw a clear line between DARPA research and research supported by other funding sources, like NSF and NIH.
One of the grad students wrote a parody of a tech report generating program that would create a new tech report given a few key words and the funding agencies. Input like "blob", "identification", "snake", "NIH" would generate a tech report on "Finding blob shape tumours using snakes" - change "NIH" to "DARPA" and the title would be "Finding blob shaped targets in side looking radar".
The overlap between ideas for miilitary applications, and non-military applications, was so extensive that you couldn't untangle them. One research student was determined not to accept military funding, and didn't, Stiill, I was in the tech report room the day the DARPA tech officer came to get a few reports, and he made a point of selecting reports by that graduate student. "Great work, gotta follow it" was his line.
So, while I fully accept and respect personal decisions not to accept military funding, any open work to advance autonomous robots will end up in weapons. That's a consequence of the research area, not the funding source. Those in this area have to enter it with no pretenses.
G.N. Hardy, in his great memoir, "Apologies of a Mathematican", proudly stated that he never worked on a military project even during the time before and during WWII. C.P. Snow, in his preface to that book, noted that while Hardy never worked on such a project, he trained all the (Cambridge?) mathematicans in the British service. Science is the ultimate open source endeavor.
In the original Apple-Apple settlement from the first suit, Apple computer agreed to stay out of music. When they added microphones and went that direction, Apple music sued again and won, but I seem to remember that Apple computer was then freer to invovate. If you care about the details you can start with
http://www.macobserver.com/news/99/april/990419/be atlesvsapple.html
Makes it hard for a 60s era kid to figure out who's the good guys in this battle of corps. I give up.
They are different things, but you're missing the relationship between the two.
When he invented the Turing machine, he included the concept of a universal Turing machine (UTM) that could simulate the behavior of any other Turing machine. You only really needed one UTM.
Since he invented the UTM, lots of people have invented other computing methods and devices (like, AMD chips). Over the years all those new methods have been proven to be no more powerful than Turing original UTM - ie, a UTM can emulate any Perl, C, Java, AMD machine code, Recursive function, or etc. Very, very slowly, yes, but a UTM gets the job done. See the Church-Turing thesis for background.
If you go back and read Turing's original 1936 article on the UTM (available online), you'll note the importance of the UTM being a state machine. Read his 1954 article on the Turing test (also online), and you'll find an echo of the logic applied to thinking about human brains.
So what Turing was asking can be viewed in part as whether a UTM could emulate the behavior of the human mind (the strong AI hypothesis). These aren't "totally different things" - there is a strong relationship. Hey, read the articles, they're accessible with a little work.
The question is whether the wetware in our brain has some computational power not possessed by a Turing machine equivalent device. If we don't have "extra" computing features, then our brains can be emulated by a Turing equivalent machine. Roger Penrose and others have attacked AI by arguing that the brain has extra computational power.
In some sense, we are talking about a simulating a human with a computer.
The iLife CD with all three is $14.95 special intro education price, so I opted for the snailmail solution. ....
Have them installed when I get home? Like I have time to use them, anyway
Bill Gates and company figured this out when Java was released. This article is much more interesting that that simple observation, though.
The key to this particular competition seems to be the size and weight limits on the cars. This is a contest for little RC-style cars. The technology exists to go much faster, but not in this form factor.
Using it to stich mosaics together.
Using it to use overlapping images to increase the effective resolution of the camera. This is called "super resolution".
Computer vision types have been doing this for years. Shmuel Peleg of Hebrew University has done some good work and had the work show up in commerical products, including Videobrush - you could take a webcam, wave it around, and in real time get a mosaic. In 1995 (I think.) Don't know if you can still get that product.
Do a google search for him and you'll find his home page and superresolution papers (Peleg and Irani is an accessible paper and one of the first - the concept they used, however, comes from Bruce Lucas' thesis at CMU.
Applications include: combining NASA satellite images of the Earth to get higher resolution, ditto for images of the human retina; and, a personal favorite, smoothing images of the system used at the Superbowl a couple of years ago where they had 75 cameras and could show any play from any angle in liive video. That was done by Takeo Kanade, Luca's' advisor.
But, I did find these supporting companies which includes Sony. There was a recent Slashdot post on how new technologies can challenge existing standards, and one claim was if the technology can find a small niche market not claimed by the big companies, it can survive - and machine vision is such a specialized, niche market.
* 3M Interconnect Solutions Division * ADIMEC Advanced Image Systems BV * Alacron, Inc. * ATMEL Grenoble * Basler Vision Technologies * BitFlow, Inc. * Cognex Corporation * Cohu, Inc., Electronics Division * Coreco Imaging * DALSA Corp * Data Translation, Inc. * Datacube, Inc. * Duncan Technologies, Inc. * Epix * Euresys S.A. Corporate Headquarters * i2S - Line Scan * illunis LLC * Intercon 1 * JAI A-S * Leutrek Vision, Inc. * Matrox Electronic Systems Ltd. * National Instruments * PerkinElmer Optoelectronics * PULNiX America, Inc. * Sony Electronics Inc. * Toshiba America Information Systems/ISD * TVI Vision * Xybion Electronic Systems
The standard is called Camera Link and you can find a technical PDF on it with a quick google search.
Camera link was designed specifically for connecting digital cameras so, according to the stuff I've seen so far, there are line scan cameras in camera link but not in firewire, camera link cameras are higher bandwidth than firewire cameras (up to 2.38 GB), and camera link is supported by the industrial (as opposed to consumer and artistic) camera and frame grabber community
I don't know much more than this, so I'm not advocating Camera Link, just putting it in this discussion for those who haven't heard of it.