I am really not certain that the current Boch radar offering is anything close that what is needed to replace a LIDAR on a self driving car. The goal is not to only detect the distance of the nearest metallic object in front of the car, but to reconstruct a spatial representation around the car precise enough to recognize and categorize any objects out there. IMHO, while today current DSP capabilities are impressive, this problem is still challenging.
What's will happens when the radars of many cars will interfere ? Even with visible light spectrum, others cars illuminations can produce situation where the perceived image is hard to interpret correctly in detail. Using a lower frequency is not likely to make the problem simpler.
While undeniably more electronics consume more power, it's not on the same scale compared to the power used by the transmission. Even 100 Watt of electronic equipment is very few to the the power need to accelerate a small car.
While I was young I was not so stressed to get my driver's licence, because I thought that I was losing times concentrating on anything passing into my head without raising risk and that self driving car will soon make driving licence useless anyway. Well, 22 years later, I finally take the time to get my driver's licence two years ago because is more comfortable with a family. But I lost my bid on the self driving car availability and lost opportunity to concentrate on new ideas on my head. Because driving still require a lot of attention trying to detect soon enough irrational situation on the road. Russian cars webcam for example show a lot of incredible situations when it's not always trivial to avoid collateral damage. In this regards I am very curious how self driving cars will handle such situations.
It' undeniable that many European (mostly man but not only) drivers are still thinking that way. Part of the problem is also the fact that historically automatic gears was more expensive, a bit less efficient, an not so well designed outside of a few (and costly) European manufactures. Now the situation is slowly changing because robotic gears with computer is efficient, easy to tune, and look like sport racer. Still too costly for the low to middle range price but like many others new features before, this will be solved over time.
The satellite gears with electromagnetic coupling of the hybrid Totota/Lexus are actually the most advanced way to solve the problem: it's simply outside of the common human capability to handle manually this gear part in real time while driving. And even if it was possible, an human will certainly fail to archive the low consumption compared to the computer.
Even if your claims will be proved right, a low speed collision is certainly not a safety problem: the energy involved is not enough to cause injury. The lost of trajectory while braking at high speed is a serious safety problem that can cause fatal accident.
The ABS is only an automated (and in fact very efficient) way to do what's professional race driver used since a long time to keep control of the trajectory. But many people forget too often that it's the tires that make all the adherence with the road and that ABS can't overcome situations where tires lost adherence.
Still, Canonical could grow into a successful company if some of there flagship projects gain large acceptance.
There is large speculation about Microsoft buying Nokia since almost two years now. Nokia was an empty bag already with Elop (no more own ecosystem, no more custom ASICs, no more factories, and collapsing sales channels) now there even have sell the bag (the Espoo HQ)...
Block access is absolutely not the problem: NFS for example uses block concurrent access for the data without any glitch.
The only pain of using USB Mass Storage is the contention when both the host and the device access the memory. It's not an excuse to restrict the users to the very limited capability of MTP. Using NFS (over IP over USB) would have been even more powerful by exposing a real full filesystem with concurrent access, multiple operations, full directory recursion, arbitrary access in both read and write mode, appending data to a file, shrinking a file, and proper caching. MTP is really horrible compared to the extremely reliable NFS.
The main problem with block device level access is that you can't access it from 2 places simultaneously, so it means the filesystem must be unmounted from the phone to be mounted on the PC.
If this is the only problem of USB Mass Storage, why not fixing it by exporting a virtual block device and internally handle files instead of blocks ? QEMU already can automatically create a virtual FAT disk image from a directory tree, albeit in read only mode yet. Write mode will be harder, but not impossible.
Look like MTP is very inferior to USB Mass Storage for a lot of users that need to do a bit more than drag and drop a file. In addition, it seem to be slow.
in a way USB mass storage is the wrong thing - it exposes the underlying device in block mode
While true for a lot of devices, this is not necessary a impossible problem. USB Mass Storage can expose a virtual filesystem that is different from the filesystem used on the physical memory, allowing to handle contention at the files level instead of at the blocks level.
I have a Nexus S updated to Jelly Bean 4.1.2 and it happily support USB Mass Storage:
lsusb: Bus 001 Device 019: ID 18d1:4e22 Google Inc. Nexus S
dmesg: [1109159.988681] usb 1-1.1.4.1: new high speed USB device using ehci_hcd and address 19 [1109160.081288] usb 1-1.1.4.1: New USB device found, idVendor=18d1, idProduct=4e22 [1109160.081293] usb 1-1.1.4.1: New USB device strings: Mfr=2, Product=3, SerialNumber=4 [1109160.081295] usb 1-1.1.4.1: Product: Nexus S [1109160.081297] usb 1-1.1.4.1: Manufacturer: samsung [1109160.081298] usb 1-1.1.4.1: SerialNumber: [1109160.081446] usb 1-1.1.4.1: configuration #1 chosen from 1 choice [1109160.083930] scsi12 : SCSI emulation for USB Mass Storage devices [1109160.084139] usb-storage: device found at 19 [1109160.084141] usb-storage: waiting for device to settle before scanning [1109165.084680] usb-storage: device scan complete [1109165.085272] scsi 12:0:0:0: Direct-Access Google File-CD Gadget 0000 PQ: 0 ANSI: 2 [1109165.088506] sd 12:0:0:0: [sde] Attached SCSI removable disk [1109274.036890] sd 12:0:0:0: [sde] 27957215 512-byte logical blocks: (14.3 GB/13.3 GiB) [1109274.037598] sd 12:0:0:0: [sde] Assuming drive cache: write through
I see two possible future for Ubuntu mobile using Android kernel:
1) Google is happy with this and allow manufacturers to release Ubuntu phones on same hardware released for Android phones. While this can look a bit odd for Google at first, this is not necessary bad for them. To be pragmatic, no OS will ever gain a total monopoly. So it's better to live with a friendly alternative to push higher pressure on the unfriendly alternatives. I suspect that a such move - to be friendly - will be extremely positive in the actual context where each OS fight each against the others. For the manufacturers, this is the same hardware, so this do not change so much there business.
2) Google is unhappy with this and he will quickly find a way to prevent Ubuntu phone install or make it so unpleasant that only a few will do it. Ubuntu phone will then be an another phone OS in the growing list of phone OS that have tried to gain acceptance but will finally failed.
I hope that Google will soon publicly announce 1). This will be a major move.
99.99999% of all Slashdot readers that claim open source is superior to closed source have not contributed one line of code to any open source project. What's the point then other then a smug superiority complex.
99.99999% of readers that claim closed source is superior to open source have not contributed one line of code to any closed source project.
Regardless of the political system, the money is not a prefect way to represent any values in the life. What kind of frustration could possibly need so much money to be satisfied ?
If you look closely into the possible responses, you will certainly notice, that in a way or in an others, owning a lot of money bring advantage only if there is a lot of others without that much money. For a precise metric, there is no advantage if there is no frustration for someone else. Now the problem is to focus everything on a single precise metric: money. A good diversity of ways of enjoying life helps everyone to shine where he is good, and to makes of the frustration of not being so good on others subject more acceptable.
Getting caught breaking a law (which is what patent infringement is) is ALWAYS considerably more expensive than doing something legally.
And what would have been the legal way in this case ? Contacting any patent holders that have claim anything possibly applying to your product, even if there exists predated works that invalid the patent, and then asking for an arbitrary price ti use it ? Seriously ?
The positive uses of patent is to make ingenious solutions available to the maximum of entities that need it and thus lowering the price to uses the patent to a small price for each entities.
The negative uses of patent is to hide claims for some years waiting for someone using something similar into a successful product and then asking them to pay a insane price to uses your patent that there never have heard before.
The positive uses case is a way to spread innovation in the wild, fast, and at low price. The negative uses case is a way to prevent innovation spread because of the risk of too high price and legal action that take time and cost a lot.
This is way so many claim that the problem is not so much the patent by itself, but the patent system that is actually making all the industries less innovative and less competitive, witch is exactly the opposite goal of the original patent system.
Agree. If you need performance and durability, you will certainly not looking at the cheapest possible memory.
In my experience, USB memory and SD Card are very reliable. I have user a lot of them and have yet to see a defective one. Even my oldest 64MB USB memory (yes MB, not GB) still work like a charm. On embedded systems I have see some memory issues, but there was related to design errors (bus or power supply noise, wrong voltage or timing, not enough current, missing protection and software weakness or bugs). There exists a couple of "Industrial grade" memories with specialized controllers that make extra checks and errors corrections.
Why did you needs an adapter ? If your target is a PC, you certainly can use a USB memory, and if your target is a embedded system, you can certainly use the SD Card in the SPI mode. I have used the two methods many times without major issue (BIOS might need an update, especially on Gigabyte mainboard).
Slashdot Mirror
I am really not certain that the current Boch radar offering is anything close that what is needed to replace a LIDAR on a self driving car. The goal is not to only detect the distance of the nearest metallic object in front of the car, but to reconstruct a spatial representation around the car precise enough to recognize and categorize any objects out there. IMHO, while today current DSP capabilities are impressive, this problem is still challenging.
Use a horse, it's certainly not as boring as a car.
What's will happens when the radars of many cars will interfere ? Even with visible light spectrum, others cars illuminations can produce situation where the perceived image is hard to interpret correctly in detail. Using a lower frequency is not likely to make the problem simpler.
While undeniably more electronics consume more power, it's not on the same scale compared to the power used by the transmission. Even 100 Watt of electronic equipment is very few to the the power need to accelerate a small car.
Mod parent up.
While I was young I was not so stressed to get my driver's licence, because I thought that I was losing times concentrating on anything passing into my head without raising risk and that self driving car will soon make driving licence useless anyway. Well, 22 years later, I finally take the time to get my driver's licence two years ago because is more comfortable with a family. But I lost my bid on the self driving car availability and lost opportunity to concentrate on new ideas on my head. Because driving still require a lot of attention trying to detect soon enough irrational situation on the road. Russian cars webcam for example show a lot of incredible situations when it's not always trivial to avoid collateral damage. In this regards I am very curious how self driving cars will handle such situations.
It' undeniable that many European (mostly man but not only) drivers are still thinking that way. Part of the problem is also the fact that historically automatic gears was more expensive, a bit less efficient, an not so well designed outside of a few (and costly) European manufactures. Now the situation is slowly changing because robotic gears with computer is efficient, easy to tune, and look like sport racer. Still too costly for the low to middle range price but like many others new features before, this will be solved over time.
The satellite gears with electromagnetic coupling of the hybrid Totota/Lexus are actually the most advanced way to solve the problem: it's simply outside of the common human capability to handle manually this gear part in real time while driving. And even if it was possible, an human will certainly fail to archive the low consumption compared to the computer.
Even if your claims will be proved right, a low speed collision is certainly not a safety problem: the energy involved is not enough to cause injury. The lost of trajectory while braking at high speed is a serious safety problem that can cause fatal accident.
The ABS is only an automated (and in fact very efficient) way to do what's professional race driver used since a long time to keep control of the trajectory. But many people forget too often that it's the tires that make all the adherence with the road and that ABS can't overcome situations where tires lost adherence.
Of course this was a joke.
Still, Canonical could grow into a successful company if some of there flagship projects gain large acceptance.
There is large speculation about Microsoft buying Nokia since almost two years now. Nokia was an empty bag already with Elop (no more own ecosystem, no more custom ASICs, no more factories, and collapsing sales channels) now there even have sell the bag (the Espoo HQ) ...
Block access is absolutely not the problem: NFS for example uses block concurrent access for the data without any glitch.
The only pain of using USB Mass Storage is the contention when both the host and the device access the memory. It's not an excuse to restrict the users to the very limited capability of MTP. Using NFS (over IP over USB) would have been even more powerful by exposing a real full filesystem with concurrent access, multiple operations, full directory recursion, arbitrary access in both read and write mode, appending data to a file, shrinking a file, and proper caching. MTP is really horrible compared to the extremely reliable NFS.
Don't panic. This is just an idea that passed in my head.
Still, I would love this to happen...
The main problem with block device level access is that you can't access it from 2 places simultaneously, so it means the filesystem must be unmounted from the phone to be mounted on the PC.
If this is the only problem of USB Mass Storage, why not fixing it by exporting a virtual block device and internally handle files instead of blocks ? QEMU already can automatically create a virtual FAT disk image from a directory tree, albeit in read only mode yet. Write mode will be harder, but not impossible.
Did't know that. I searched more on that and found harsh comments about MTP, like on this forum:
http://androidforums.com/motorola-droid-razr-hd-maxx-hd/658716-jelly-bean-usb-mass-storage.html
Look like MTP is very inferior to USB Mass Storage for a lot of users that need to do a bit more than drag and drop a file. In addition, it seem to be slow.
How much USB memory or USB hard disk implement MTP ?
in a way USB mass storage is the wrong thing - it exposes the underlying device in block mode
While true for a lot of devices, this is not necessary a impossible problem. USB Mass Storage can expose a virtual filesystem that is different from the filesystem used on the physical memory, allowing to handle contention at the files level instead of at the blocks level.
Android Jelly Bean axes USB Mass Storage
False claim, sorry.
I have a Nexus S updated to Jelly Bean 4.1.2 and it happily support USB Mass Storage:
lsusb:
Bus 001 Device 019: ID 18d1:4e22 Google Inc. Nexus S
dmesg:
[1109159.988681] usb 1-1.1.4.1: new high speed USB device using ehci_hcd and address 19
[1109160.081288] usb 1-1.1.4.1: New USB device found, idVendor=18d1, idProduct=4e22
[1109160.081293] usb 1-1.1.4.1: New USB device strings: Mfr=2, Product=3, SerialNumber=4
[1109160.081295] usb 1-1.1.4.1: Product: Nexus S
[1109160.081297] usb 1-1.1.4.1: Manufacturer: samsung
[1109160.081298] usb 1-1.1.4.1: SerialNumber:
[1109160.081446] usb 1-1.1.4.1: configuration #1 chosen from 1 choice
[1109160.083930] scsi12 : SCSI emulation for USB Mass Storage devices
[1109160.084139] usb-storage: device found at 19
[1109160.084141] usb-storage: waiting for device to settle before scanning
[1109165.084680] usb-storage: device scan complete
[1109165.085272] scsi 12:0:0:0: Direct-Access Google File-CD Gadget 0000 PQ: 0 ANSI: 2
[1109165.088506] sd 12:0:0:0: [sde] Attached SCSI removable disk
[1109274.036890] sd 12:0:0:0: [sde] 27957215 512-byte logical blocks: (14.3 GB/13.3 GiB)
[1109274.037598] sd 12:0:0:0: [sde] Assuming drive cache: write through
USB mass storage is an anachronysm.
If you never have see a USB memory or USB hard disk, then you can define yourself as a anachronism.
I see two possible future for Ubuntu mobile using Android kernel:
1) Google is happy with this and allow manufacturers to release Ubuntu phones on same hardware released for Android phones. While this can look a bit odd for Google at first, this is not necessary bad for them. To be pragmatic, no OS will ever gain a total monopoly. So it's better to live with a friendly alternative to push higher pressure on the unfriendly alternatives. I suspect that a such move - to be friendly - will be extremely positive in the actual context where each OS fight each against the others. For the manufacturers, this is the same hardware, so this do not change so much there business.
2) Google is unhappy with this and he will quickly find a way to prevent Ubuntu phone install or make it so unpleasant that only a few will do it. Ubuntu phone will then be an another phone OS in the growing list of phone OS that have tried to gain acceptance but will finally failed.
I hope that Google will soon publicly announce 1). This will be a major move.
99.99999% of all Slashdot readers that claim open source is superior to closed source have not contributed one line of code to any open source project. What's the point then other then a smug superiority complex.
99.99999% of readers that claim closed source is superior to open source have not contributed one line of code to any closed source project.
What your point really ?
Interesting philosophic question: could the success only be the result of breaking rules of a systems ?
Nothing political in my post, sorry.
Regardless of the political system, the money is not a prefect way to represent any values in the life. What kind of frustration could possibly need so much money to be satisfied ?
If you look closely into the possible responses, you will certainly notice, that in a way or in an others, owning a lot of money bring advantage only if there is a lot of others without that much money. For a precise metric, there is no advantage if there is no frustration for someone else. Now the problem is to focus everything on a single precise metric: money. A good diversity of ways of enjoying life helps everyone to shine where he is good, and to makes of the frustration of not being so good on others subject more acceptable.
Getting caught breaking a law (which is what patent infringement is) is ALWAYS considerably more expensive than doing something legally.
And what would have been the legal way in this case ? Contacting any patent holders that have claim anything possibly applying to your product, even if there exists predated works that invalid the patent, and then asking for an arbitrary price ti use it ? Seriously ?
The positive uses of patent is to make ingenious solutions available to the maximum of entities that need it and thus lowering the price to uses the patent to a small price for each entities.
The negative uses of patent is to hide claims for some years waiting for someone using something similar into a successful product and then asking them to pay a insane price to uses your patent that there never have heard before.
The positive uses case is a way to spread innovation in the wild, fast, and at low price. The negative uses case is a way to prevent innovation spread because of the risk of too high price and legal action that take time and cost a lot.
This is way so many claim that the problem is not so much the patent by itself, but the patent system that is actually making all the industries less innovative and less competitive, witch is exactly the opposite goal of the original patent system.
Maybe still not enough to trigger any reaction ?
We will soon live in a world without any privacy, paying for everything, and where thinking is forbidden.
Money, money, money....
Still something to eat ?
Agree. If you need performance and durability, you will certainly not looking at the cheapest possible memory.
In my experience, USB memory and SD Card are very reliable. I have user a lot of them and have yet to see a defective one. Even my oldest 64MB USB memory (yes MB, not GB) still work like a charm. On embedded systems I have see some memory issues, but there was related to design errors (bus or power supply noise, wrong voltage or timing, not enough current, missing protection and software weakness or bugs). There exists a couple of "Industrial grade" memories with specialized controllers that make extra checks and errors corrections.
Why did you needs an adapter ? If your target is a PC, you certainly can use a USB memory, and if your target is a embedded system, you can certainly use the SD Card in the SPI mode. I have used the two methods many times without major issue (BIOS might need an update, especially on Gigabyte mainboard).
Let me know when I can get a 32GB SSD for $20 or less.
Actually, the cheapest 32GB SDCard or USB memory are about that price.
Ok, there are not as fast as a SSD, but still show a valid indication that the price per capacity can be improved in the short term.