This is not a patent on autonomous driving. That would be very hard to do given the extensive prior art. Instead, the patent is basically focused the vehicle switching modes or executing an autonomous motion based on a QR code or some other infrastructure based marker that points at an URL. Many other teams have used infrastructure markers to indicated a change in vehicle mode. For example, California PATH encodes a binary 0101 signal using N-S-N-S magnet orientation, several Japanese teams use RF-based roadway beacons, and a variety of teams use painted markings (e.g., Civis bus in Las Vegas). However, most of these use internally stored references and maps on the vehicles. Also, some of the DSRC implementations conceived by US DOT include autonomous actions based on information received over RF from nearby vehicles. The difference here is that this patent is about using the reference to look up a command over the internet. It is a small delta on existing work.
Having said this, the idea raises all sorts of questions about man-in-the-middle attacks.
These book chains are dying because they're trying to do business as if nothing has changed. They're hiring the cheapest, dumbest possible labour when people are only willing to go to a bookstore and pay a bit more than they would at Amazon because they want to talk to someone knowledgeable and well-read about books.
Actually, the flagship Borders in Ann Arbor was notoriously difficult to get hired at. They had a hard core literature test for prospective employees. I know English majors who failed the test.
Everyone is looking at the tree, not the forest. While everyone is going to jump on the "Apple did this to make money" argument, you know a major reason for this key was Apple's way of keeping content providers happy. Now that it's broken, there is a new "analog hole" for audio and video content. It is easy to imagine a computer using this to create a digital media file rather than routing to speakers. I suspect it won't be long before content providers pressure Apple into using secondary data to confirm iTunes is talking to a legit device.
Google actually built an all-star team. They hired the tech leads for both the Carnegie Mellon team (top finisher for 1st Grand Challenge and won the Urban Grand Challenge) and the Stanford team (won the 2nd Grand Challenge). They also hired a bunch of the other developers from each team. I was at Chris Urmson's recent presentation at Carnegie Mellon about the Google cars. The approach the Google team is using looks predominately like the CMU method but with key features of the Stanford team.
Some well known facts within the robotics community, but not outside: The Stanford team was a former CMU prof (Thrun) and his technical lead (Montemerlo) was still finishing his CMU PhD while working for Stanford. His co-advisor was the prof running the CMU team (Whittaker) and his CMU classmate was the technical lead for CMU's teams (Urmson). Google hired three out of four on this list.
Yep, and autonomous driving in mixed traffic was done for No Hands Across America in 1995. Also, SARTRE is very similar to the CHAUFFEUR I and II European Commission projects in the 90's and early 00's.
Urmson (PhD, faculty on leave), Montemerlo (PhD), and Thrun (former faculty) all have ties to Carnegie Mellon. Autonomous driving has been a steady effort at CMU. For example, No Hands Across America was in 1995.
However, I'd argue that 3D movies have already gotten past the blue LED phase. Certainly Cameron's Avatar was a highly engrossing (both to the viewer and the bottom line) film even without the 3D, and without throwing somebody's yo-yo in your face...
I've been telling people that Cameron got Avatar "right" in terms of 3D exactly for this reason. There is such a stark contrast between it and other 3D movies in that there were only a couple scenes where it was clear they were showing off the 3D. Even those had reasons where the scene kind of made sense (like refocusing on near/far during the diary videos). I think Avatar will be a real benchmark in 3D strictly because it shows you can do well with 3D without being an eye-poker movie. It will be interesting to see how many other directors learn from Cameron's willingness to try to do it right.
Don't post media on unprotected pages. No big loss behind this step. Friends and family can handle a simple user/password combination - we've been doing this for years. Trust me, the rest of the world doesn't really want to see your pictures of the kids at their friend Joey's birthday party.
Grandparent: There is a pretty big difference between CC and subtitles. The former can be manipulated in format, font, etc and often include details like "Phone Ringing" that are left out of subtitles. Even subtitles that are supposedly for the deaf and hard of hearing often lack these audible descriptors. Subtitles also don't move based on the scene.. For example, text for edited CC is often positioned from side to side to match two people bantering or italicized for an off screen speaker (example). Real-time CC, which is not typical in movies, is the stuff you see scrolling at the top or bottom of screens in bars during sports and news. That's closer to subtitles due to the time pressure the captioner is under.
Parent: There is a digital CC spec but, like line21, it basically gets hosed by HDMI.
I can live with confusing names if they get around to supporting closed captioning data like they are supposed to. They misinterpreted the legal requirements for closed captioning as it being something which is handled by set-top boxes rather than TVs and elected to not transmit the data. HDMI's own FAQ makes this position clear. However, the law is quite clear that the TVs are required to render captions. Unfortunately, people use devices other than set-top boxes to push content to the TV. If you need captioning, you can't use HDMI with Blu-ray disc players or other devices.
There is some credence to this line of thought. For years states like Arizona had the highest fatality rates. This was attributed to the very straight highways where people could put the pedal down and barely steer. Speed doesn't kill, but delta v does. Crashing from 90 mph is a lot worse than 65 mph.
Actually, the old TRS-80 Model 100 from the early 80's ran on AA batteries. There was some merit to being able to run down to the corner store when you were in a bind. Yes, I used to use one of these. It was a damn cool computer at the time.
Actually, joystick drive-by-wire systems have been on the market for some time now. These systems are often installed for people with disabilities who cannot manipulate a regular steering wheel. The vehicle modification industry is a mature market with its own national association.
This concept vehicle addresses one of the core problems - cost. A typical vehicle conversion can run well upwards of $20k (in addition to vehicle cost) and involves massive hardhacking of the vehicle itself. You're pretty much stuck with the vehicle until it dies since you won't be able to sell it. It is far better to have an OEM route.
In terms of safety and control dynamics, I'm pretty comfortable that these are surmountable issues.
First, most people who currently get a joystick system receive dedicated training by a certified driver rehabilitation therapist (CDRS). I would imagine that lessons learned in this field could be distilled down into a more generalized training approach for populations that don't have the severe dexterity impairments present in the current user base.
Second, the years of experience in the vehicle modification business is a good start towards safe control dynamics. Advanced vehicle control systems are the next big leap and are far better than they were even a decade ago. Paired with systems like lane tracking, stability control, and forward ranging systems, it would be quite possible to put a layer of "smarts" over the command inputs.
Third, removing the steering column makes it a lot easier to protect the driver in the event of a crash.
Mac mini, bluetooth keyboard and mouse. It is small, doesn't look ugly under your TV, has a super quiet fan, and you can get plenty of video adapters for whatever TV you have. It also has a DVD drive, so you can toss your DVD player. You can even get an EyeTV USB-stick add-on for DVR capability and export capability to your iPod/PMP. If you really want, you can even run a long USB extension cable to your couch so you can plug in a joystick and play video games. Likewise, you can also set it up as a home media server and/or remote access gateway when you're out and about.
Basically, you can do just about anything with one box.
This is probably outside the price range of most folks, but BodyMedia makes a research version.
And before you complain, yes research versions of such equipment are almost always more expensive than consumer versions. This has to do with the added technical support ("we want people to do [insert crazy unusual thing] while using your device, will it work?") and typical "hey, that's odd data, can you explain it?" types of follow-up. When you're doing research, this level of support and debugging has a definite, non-zero value.
Marines: I expect a good chunk of your R&D budget for this design.
The base R&D has been done. They clearly say in the article they want something like Dragon Runner with more capability.
I know a bit about Dragon Runner. Trust me, it's seriously cool and very well engineered. If you don't believe it has the "throwable" part down, watch this movie.
Slashdot Mirror
This is not a patent on autonomous driving. That would be very hard to do given the extensive prior art. Instead, the patent is basically focused the vehicle switching modes or executing an autonomous motion based on a QR code or some other infrastructure based marker that points at an URL. Many other teams have used infrastructure markers to indicated a change in vehicle mode. For example, California PATH encodes a binary 0101 signal using N-S-N-S magnet orientation, several Japanese teams use RF-based roadway beacons, and a variety of teams use painted markings (e.g., Civis bus in Las Vegas). However, most of these use internally stored references and maps on the vehicles. Also, some of the DSRC implementations conceived by US DOT include autonomous actions based on information received over RF from nearby vehicles. The difference here is that this patent is about using the reference to look up a command over the internet. It is a small delta on existing work.
Having said this, the idea raises all sorts of questions about man-in-the-middle attacks.
These book chains are dying because they're trying to do business as if nothing has changed. They're hiring the cheapest, dumbest possible labour when people are only willing to go to a bookstore and pay a bit more than they would at Amazon because they want to talk to someone knowledgeable and well-read about books.
Actually, the flagship Borders in Ann Arbor was notoriously difficult to get hired at. They had a hard core literature test for prospective employees. I know English majors who failed the test.
I got crumbs on my keyboard...
Everyone is looking at the tree, not the forest. While everyone is going to jump on the "Apple did this to make money" argument, you know a major reason for this key was Apple's way of keeping content providers happy. Now that it's broken, there is a new "analog hole" for audio and video content. It is easy to imagine a computer using this to create a digital media file rather than routing to speakers. I suspect it won't be long before content providers pressure Apple into using secondary data to confirm iTunes is talking to a legit device.
Google actually built an all-star team. They hired the tech leads for both the Carnegie Mellon team (top finisher for 1st Grand Challenge and won the Urban Grand Challenge) and the Stanford team (won the 2nd Grand Challenge). They also hired a bunch of the other developers from each team. I was at Chris Urmson's recent presentation at Carnegie Mellon about the Google cars. The approach the Google team is using looks predominately like the CMU method but with key features of the Stanford team.
Some well known facts within the robotics community, but not outside: The Stanford team was a former CMU prof (Thrun) and his technical lead (Montemerlo) was still finishing his CMU PhD while working for Stanford. His co-advisor was the prof running the CMU team (Whittaker) and his CMU classmate was the technical lead for CMU's teams (Urmson). Google hired three out of four on this list.
3, 2, 1...
Yep, and autonomous driving in mixed traffic was done for No Hands Across America in 1995. Also, SARTRE is very similar to the CHAUFFEUR I and II European Commission projects in the 90's and early 00's.
The AHS History 1939 - 1997 movie (RealPlayer) is awesome.
Clearly, GP is not familiar with the Mother of All Demos. The main computer was on the other end of a 1968 modem.
Urmson (PhD, faculty on leave), Montemerlo (PhD), and Thrun (former faculty) all have ties to Carnegie Mellon. Autonomous driving has been a steady effort at CMU. For example, No Hands Across America was in 1995.
Mmm... Duff's.
Here is a comment from someone with a higher digit number telling a lower digit person "you must be new here"
However, I'd argue that 3D movies have already gotten past the blue LED phase. Certainly Cameron's Avatar was a highly engrossing (both to the viewer and the bottom line) film even without the 3D, and without throwing somebody's yo-yo in your face ...
I've been telling people that Cameron got Avatar "right" in terms of 3D exactly for this reason. There is such a stark contrast between it and other 3D movies in that there were only a couple scenes where it was clear they were showing off the 3D. Even those had reasons where the scene kind of made sense (like refocusing on near/far during the diary videos). I think Avatar will be a real benchmark in 3D strictly because it shows you can do well with 3D without being an eye-poker movie. It will be interesting to see how many other directors learn from Cameron's willingness to try to do it right.
Don't post media on unprotected pages. No big loss behind this step. Friends and family can handle a simple user/password combination - we've been doing this for years. Trust me, the rest of the world doesn't really want to see your pictures of the kids at their friend Joey's birthday party.
Grandparent: There is a pretty big difference between CC and subtitles. The former can be manipulated in format, font, etc and often include details like "Phone Ringing" that are left out of subtitles. Even subtitles that are supposedly for the deaf and hard of hearing often lack these audible descriptors. Subtitles also don't move based on the scene.. For example, text for edited CC is often positioned from side to side to match two people bantering or italicized for an off screen speaker (example). Real-time CC, which is not typical in movies, is the stuff you see scrolling at the top or bottom of screens in bars during sports and news. That's closer to subtitles due to the time pressure the captioner is under.
Parent: There is a digital CC spec but, like line21, it basically gets hosed by HDMI.
I can live with confusing names if they get around to supporting closed captioning data like they are supposed to. They misinterpreted the legal requirements for closed captioning as it being something which is handled by set-top boxes rather than TVs and elected to not transmit the data. HDMI's own FAQ makes this position clear. However, the law is quite clear that the TVs are required to render captions. Unfortunately, people use devices other than set-top boxes to push content to the TV. If you need captioning, you can't use HDMI with Blu-ray disc players or other devices.
Perhaps it would be less confusing if translated into Swedish Chef.
There is some credence to this line of thought. For years states like Arizona had the highest fatality rates. This was attributed to the very straight highways where people could put the pedal down and barely steer. Speed doesn't kill, but delta v does. Crashing from 90 mph is a lot worse than 65 mph.
AAs are a horrible way to power a laptop.
Actually, the old TRS-80 Model 100 from the early 80's ran on AA batteries. There was some merit to being able to run down to the corner store when you were in a bind. Yes, I used to use one of these. It was a damn cool computer at the time.
I'm all in favor of something that does the job better than HDMI. If you need closed captioning in the US, HDMI doesn't work for you. It omits the necessary data from the video source.
This concept vehicle addresses one of the core problems - cost. A typical vehicle conversion can run well upwards of $20k (in addition to vehicle cost) and involves massive hardhacking of the vehicle itself. You're pretty much stuck with the vehicle until it dies since you won't be able to sell it. It is far better to have an OEM route.
In terms of safety and control dynamics, I'm pretty comfortable that these are surmountable issues.
Mac mini, bluetooth keyboard and mouse. It is small, doesn't look ugly under your TV, has a super quiet fan, and you can get plenty of video adapters for whatever TV you have. It also has a DVD drive, so you can toss your DVD player. You can even get an EyeTV USB-stick add-on for DVR capability and export capability to your iPod/PMP. If you really want, you can even run a long USB extension cable to your couch so you can plug in a joystick and play video games. Likewise, you can also set it up as a home media server and/or remote access gateway when you're out and about.
Basically, you can do just about anything with one box.
This is probably outside the price range of most folks, but BodyMedia makes a research version.
And before you complain, yes research versions of such equipment are almost always more expensive than consumer versions. This has to do with the added technical support ("we want people to do [insert crazy unusual thing] while using your device, will it work?") and typical "hey, that's odd data, can you explain it?" types of follow-up. When you're doing research, this level of support and debugging has a definite, non-zero value.
Is this what they look like?
The base R&D has been done. They clearly say in the article they want something like Dragon Runner with more capability.
I know a bit about Dragon Runner. Trust me, it's seriously cool and very well engineered. If you don't believe it has the "throwable" part down, watch this movie.
The 90's