Separate names with a comma.
Discussion in 'The BS Topic' started by bourbon_scotch, Nov 8, 2017.
Wait till hackers from have fun with you in your new $100,000 HAL car
I actually just read an article on just that. It's really not that hard to secure a car's ECM's; much easier than a conventional computer.
Since the ECM's generally only perform very specific tasks - it's very easy to monitor what they are up to and limit access.
Like it or not, autonomous cars are happening
FYI - Waymo = Google. They have really good people, and all of the money in the world to make this work.
Last I heard one of the most difficult technical challenges right now is snow. The systems maintain lane alignment based on the edge stripes, and during a snow squall those tend to disappear, sometimes very quickly. Not sure what they have in mind to deal with that.
A lot of us in the Detroit auto industry think Lutz is right.
If self driving cars can reduce auto deaths from 30,000+ a year to 5,000, or less, I think it's inevitable in the coming decades that people in general will not be allowed to drive.
There may be exceptions in the very rural areas, where roads are very rough; and they'll probably allow some recreation, like off-roading and racing. But large cities will mostly ban cars.
Several years ago, a hacker showed how to do just this. He controlled the accelerator, brakes, transmission shifting and even opened the windows on a Jeep. This was before the advent of electric steering but I'll bet that will be a possibility soon.
I suspect the US population may chafe at the idea that the government says they aren't allowed to drive. Ride an old Harley (sort of a two wheeled tractor) for serious enjoyment.
As far as hacking goes, if you make it available on the internet, it will be hacked. Do you want your steering controlled by the internet?
ok now i'm scared.
Yes, I read about Hotz. I think he's overstating his case a bit.
IIRC, some computer scientists on the West Coast - University of Washington, maybe?- wrote a paper about how they took control of a car 3 or 4 years ago. The two main vulnerabilities I recall were the ALDL(diagnostic) port, and the entertainment system. Both allowed access to the vehicle's serial CAN bus, which is what the various primary control modules use to communicate. But it took them over 6 months to figure it out, and they had way-above-average knowledge of computers.
That study actually triggered the creation of cybersecurity departments in the Detroit automakers, and since then much has been done to prevent that kind of thing from happening with our vehicles. But of course it's hard to actually prove it can never happen.
It was actually 6 years ago - time flies - here's a NYT review of the research. University of Washington, San Diego, and California. I can't find the original paper I read, it's probably out there somewhere.