Slashdot Mirror
Making Driverless Cars Safer
colinneagle writes "Several autonomous cars have been developed elsewhere, most famously by Google, and they are generally capable of identifying objects in the road directly ahead of or behind them. The challenge undertaken by MIT researchers is making these cars aware of dangers lurking around corners and behind buildings. MIT PhD student Swarun Kumar showed a video of a test run by the MIT researchers in which an autonomous golf cart running the technology, called CarSpeak (PDF), encountered a pedestrian walking from the entrance of a building to a crosswalk. The golf cart stopped roughly five yards ahead of the crosswalk and waited long enough for the pedestrian to walk to the other side of the road. The vehicle then continued driving automatically. The solution Kumar presented is based on a method of communications that is intended to expand the vehicle's field of view. This can be accomplished by compressing and sharing the data that autonomous vehicles generate while they're in motion, which Kumar says can amount to gigabits per second. In a comparison test, a car using CarSpeak's MAC-based communications was able to stop with a maximum average delay of 0.45 seconds, compared to the minimum average delay time of 2.14 seconds for a car running 802.11, the report noted."
How do these things perform in weather? ex. Blizzards
I'd hate to wind up in a snow drift in the middle of the road rather than backing up and finding an alt. route... or going home.
Even with faster stopping, there will be those who deliberately jump in front of cars in order to get hit, hopefully to score a big jury verdict.
The solution -- a camera that turns on and records encounters with pedestrians, bicyclists, etc, with a timer in place. That way, if there is a wreck, there is documented proof that the other party jaywalked or violated traffic laws.
Of course, if it is the car's fault, it will be documented as well, but assuming a fully automatic vehicle which obeys all traffic signals, it likely won't be the vehicle that caused the collision.
Does not 802.11 a (wireless) Ethernet... ... thus the notion of MAC still applies?
Uh... what?
systemd is Roko's Basilisk.
Nah. The next version of the Driverless software will target bad human drivers and run them off the road, increasing safety for everyone else!
Everything is better with chainsaws.
Do not want. It's obvious at this point that the real deal with all these innovations is to retain more and more control over what people do and where they go. They entice us with convenience as they remove the control. I realize this article is about technical minutiae, but I have no desire to help this project along.. Until society matures such that those in charge don't have insatiable desires to micromanage individual choice as much as possible, I'd rather deal with driving my own vehicles around, thanks. Besides, with the right fit, driving a car is enjoyable.
So if the driverless car stops in the road for a perceived pedestrian that may-or-may not be crossing the street will it give me the electronic finger when I lean on the horn?
Seriously, I see a lot of people standing close to the edge of the sidewalk that I think might be going to cross. Usually it turns out they are just chatting and aren't going anywhere. I suspect there will be a lot of false positives resulting in the driverless car slowing or stopping in traffic for someone who isn't actually crossing the street.
I can also anticipate kids having fun with this by "faking out" the autonomous vehicles for a laugh.
From the OP: "a car using CarSpeak's MAC-based communications was able to stop with a maximum average delay of 0.45 seconds"
This acronym 'MAC' is not used or explained anywhere else in TFS, so it's unclear whether they mean Media Access Control from the IEEE 802 spec (which probably is employed in moving data wirelessly from car to car, but has little to do with the specific problem of detecting or responding to safety hazards) or something else entirely.
DRM: Terminator crops for your mind!
Yes, and this communication is not authenticated... which means you now have up to seven tons of machinery barreling around a corner... and if it's told that the way is clear, instead of blocked, instead of a gentle deceleration and safe crossing you get human hamburger. Up next on CSI... hacking GPS signals and inter-car communication to create the perfect murder: No forensic evidence, looks just like an accident.
I do not like the idea of autonomous cars depending on or accepting unauthenticated inputs, or having two-way communication abilities while in operation. We already have a pile of broken nuclear facilities in Iran caused entirely by malicious digital communications, the source of which can't be proven. Most systems rely on GPS and network communication for route planning, which is problematic enough but can probably be made reasonably secure... but when you start processing realtime data from unauthenticated sources to make operating decisions, not just navigation decisions, I just don't see it as being possible to secure because of the wide number of variables which could be influenced independently or collectively to create an unsafe condition.
#fuckbeta #iamslashdot #dicemustdie
I checked TFA, and found this:
As noted in this report on the project, standard 802.11 networks cannot accommodate the data transmission needs for communication between autonomous vehicles because they generate more data than the available bandwidth can handle. CarSpeak instead uses a content-centric MAC protocol for transmitting data, in which data pertaining to specifically requested roads and regions contends for space in the medium, as opposed to the cars sending requests for information. This ensures the network only displays relevant data, avoiding a flood of data pertaining to open roads.
So, yes, apparently they're talking about low-level wireless networking protocols, but... it's like saying that your revolutionary new Web search engine is "copper-based." I mean, that's what the conductors on the server CPU are made of, and without copper none of it would work, but it hardly captures what's unique or noteworthy about the technology.
DRM: Terminator crops for your mind!
When people mention how autonomous vehicles can share information with each other, they implicitly assume that the vehicles and other entities within the environment will play fair and honest.
What happens if any of those systems are hacked either for nefarious reasons or just so that the driver of the hacked car can gain some advantage by sharing misinformation. ?
In this setup of autonomous vehicles, they become essentially computers on wheels. The issues that are faced in network security can manifest themselves with autonomous vehicles.
Theoretically there is always a chance of a crash. I would expect the chances of crash to be less than human driven cars and would expect insurance to be way cheaper. But yes, the car would need insurance. And yes, it would be the cars fault, and insurance premium for all cars of the model would increase.
Why do we need driverless cars?
The largest use of automobiles is to transport the driver (the sole occupant) around a city, or between cities.
So if you take the driver out of the vehicle, why does it need to go anywhere?
From the OP: "a car using CarSpeak's MAC-based communications was able to stop with a maximum average delay of 0.45 seconds"
This acronym 'MAC' is not used or explained anywhere else in TFS, so it's unclear whether they mean Media Access Control from the IEEE 802 spec (which probably is employed in moving data wirelessly from car to car, but has little to do with the specific problem of detecting or responding to safety hazards) or something else entirely.
They explain MAC right in the paper ( which is linked in the article ). It's MAC just like you think MAC is (Media Access Control).
Really, the gist of the paper, is instead of each car being the source or identity of a packet, via normal MAC 'addressing' and trying to communicate some important information ( like soft squish target...er.. human at X,Y,Z moving Z-> Y-> Z-> at such and such a rate ) via the full OSI model ( like packaging that info in UDP or TCP), You instead break down the 3D space around the car ( and other cars do the same thing ) using an octree graph ( just like visibility systems in 3D game engines), and send out this info with the MAC layer altered to show which region of the octree your information is pertinent too.
So if you are Car A, and Car B,C,D,E,F,G,H,I are all in your broadcast range, and they are dumping out gigabytes of network aware info based on their laser scanners, you can quickly and at a very low level (hardware) pick out the packets that are important to you (from the air).
tl;dr:
It's a clustered index for wireless packets based on GPS location of events stuffed into the MAC (data link) layer. It's a complicated QOS scheme that has been crafted around a specific engineering latency problem.
-Malakai
A Dragon Lives in my Garage
I have just realised the major problem with the driverless car !! People knowing their behaviour and hacking it. In other words the cars are being developed to emulate how humans drive responding to normally expected behaviour of other humans. The problem that humans will treat driverless cars differently. For example a human realising that it is a driverless car will cut in front of it knowing it will handle it. A pedestrian will step right in front of it and then step back, or will pretend to walk to the edge of the pavement ... and stop. This is what I can think of now. "Hackers" will understand how driverless cars will behave to external inputs and exploit that behaviour much like hackers exploit computer systems everywhere. It may take much longer to deal with that than it is to develop a safe driverless car in "normally predictable" scenarios.