College Senior Turns His Honda Civic Into a Self-Driving Car Using Free Hardware, Software (technologyreview.com) 132
holy_calamity writes: University of Nebraska student Brevan Jorgenson swapped the rear-view mirror in his 2016 Honda Civic for a home-built device called a Neo, which can steer the vehicle and follow traffic on the highway. Jorgenson used hardware designs and open-source software released by Comma, a self-driving car startup that decided to give away its technology for free last year after receiving a letter asking questions about its functionality from the National Highway Traffic Safety Administration (NHTSA). Jorgenson is just one person in a new hacker community trying to upgrade their cars using Comma's technology. "A Neo is built from a OnePlus 3 smartphone equipped with Comma's now-free Openpilot software, a circuit board that connects the device to the car's electronics, and a 3-D-printed case," reports MIT Technology Review. The report notes that Neodriven, a startup based in Los Angeles, has recently started selling a pre-built Neo device that works with Comma's Openpilot software, but it costs $1,495.
Re: All the better to 'drive' stoned (Score:2)
Of course you can drive stoned. It will impair your reaction time a bit, but to be fair... It's probably still better than most older drivers can manage.
Re: All the better to 'drive' stoned (Score:2)
Of course you can drive stoned
Unless you're "one toke over the line" and suffer a panic attack. Well, even then a salty stoner can still get up and down the highway, but it is by no means pleasant.
Re: (Score:2)
Of course you can drive stoned
Unless you're "one toke over the line" and suffer a panic attack. Well, even then a salty stoner can still get up and down the highway, but it is by no means pleasant.
Dude! You just need to drive 5 miles an hour so you don't attract attention...
Re: (Score:2)
Actually, that's why stoned drivers tend to be OK but drunk drivers are a problem. The drunk driver is over-confident in their abilities and tends to crash. Stoned drivers are generally more capable than they think they are and slow down more than enough to compensate for their poor reaction time.
Re: (Score:2)
Of course you can drive stoned
Unless you're "one toke over the line" and suffer a panic attack.
Sweet (cellphone zombie) Jesus!
Re: (Score:2)
Of course you can drive stoned. It will impair your reaction time a bit, but to be fair... It's probably still better than most older drivers can manage.
This is the cold hard truth the authorities will never admit. The eligibility for driving should be purely based on merit and that's it. If you can pass the test stoned, drunk or asleep then it's still a pass. If you don't like that then make the test tougher.
Re: (Score:2)
Agreed on the stoned part. Limited experiments on this suggest that what you lose in ability, you more than make up for in caution.
However, the part about older drivers is ignorant. Older drivers have lower insurance mostly because they have less accidents than young drivers. Older drivers are objectively better than young ones. Though of course there is an upper limit, where senility sets in.
Re: (Score:2)
Older drivers have lower insurance mostly because they have less accidents than young drivers. Older drivers are objectively better than young ones. Though of course there is an upper limit, where senility sets in.
They might be safer than kids, but they're still on a bunch of medications on which they really shouldn't be driving. And in a crisis, their reflexes are typically awful, often for the same reason.
Re: (Score:2)
They might be safer than kids, but they're still on a bunch of medications on which they really shouldn't be driving. And in a crisis, their reflexes are typically awful, often for the same reason.
Crisis? What crisis?
That's a huge benefit of being an experienced driver - you smell trouble far before it happens. You know which drivers are going to do something stupid, and you stay out of their way. You leave others room to make mistakes and recover from it.
A good driver will avoid many situations where good reflexes would be needed, In addition, a good driver will make sure they have enough space to not act by blind reflex, but by thinking. Your "fast reflexes" will quite often just make the sit
Re: (Score:2)
Crisis? What crisis?
I live on Clear Lake in California. We have a shitload of accidents here, often involving fatalities, which are head-on or t-bone collisions. When someone comes across the line, there may not physically be enough room to brake before they cream you — going full speed. And there are areas with essentially no shoulder.
Re: (Score:2)
I had a friend that bragged about driving drunk. He did it all the time and he got away with it for years. He dodged the cops, payed them off a couple of times. Then he got in a bad accident and people nearly died. He was driving the wrong way on a 4 lane highway. I'm 57 now and I've known a lot of functional drunks. They all got progressively worse over the years as they aged. One went to work every day and drank himself to sleep every night for years. He occasionally took a week off and when he ca
Re: (Score:2)
Also, drinking slows your reaction time and coordination, in addition to your perception. While a stoner might not be paying any damn attention, his reflexes and motor skills are not actually impaired at all.
Cannabis does affect reaction time, though it doesn't harm coordination. Informal studies with video games suggest that it may actually be a performance-enhancing drug, because it takes away twitchiness. :) But what it doesn't do that alcohol does is impair your ability to say no — notably, to yourself. Alcohol lowers inhibitions in a way that THC doesn't. That's why you find people going 100 in a 65 while drunk, and 65 in a 65 when high.
Re: (Score:2)
Works for boats [diydrones.com] too.
Re: (Score:2)
It's not a problem offroad. On public streets I imagine the law might frown on it.
Re: (Score:2)
Re: (Score:2)
I don't see why, if he drove the car. On the other hand, if he sat there while the computer drove, that's a different story.
Re: (Score:2)
He should be arrested, immediately.
On what grounds?
Another BS article. (Score:3)
This software just replaced LKAS and ACC in Hondas, both features you can get as standard anyway.
basically steering lane control in cruise mode, and adaptive cruise so you dont run in to the back of the car in front.
That is far FAR from a 'self driving car', not even similar. The system cannot even operate below 18mph/25mph depending on model.
Still, new media and all, who CARES if any facts are checked, its all about the HYPE!
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
both features you can get as standard anyway
Sure and you can just get a Tesla too. But what if you don't want to buy an entire new car to get a feature?
Re: (Score:2)
From TFA:
federal and state laws probably don’t pose much of a barrier to those with a desire to upgrade their vehicle to share driving duties. NHTSA has authority over companies selling vehicles and systems used to modify them, but consumers have significant flexibility in making changes to their own vehicle, says Smith, who advises the U.S. Department of Transportation on law and automation.
Anyone using a home-built Neo will still have to comply with state rules requiring responsible driving, though. (Comma’s Openpilot software tries to help with that: it complains if the driver doesn’t touch the wheel every five minutes, and it asks for human intervention if it’s having trouble interpreting the road ahead.) And in the event of a crash, using a home-built driving aid might raise eyebrows. “Just because you can legally operate it doesn’t mean you are not civilly liable,” says Smith.
Re: (Score:2)
Re: (Score:2)
Look, if it's ok for google and co to have their half-baked crap navigate on public roads...
The difference is that Google has applied for, and been given, a license to test their SDCs on public roads.
Money doesn't bring people back from the dead.
But money does pay for repairs and medical bills. Injuries are far more common than fatalities, and non-injury accidents causing vehicle damage are even more common.
Re: (Score:2)
Google did an awful lot of testing on an AI test facility and proved basic competence before they ventured on to the public roads. As have all the other legitimate autonomous driving developers.
Re: (Score:2)
Re: (Score:1)
Re: Interesting observation. (Score:1)
Why are you thinking about boys?
Re: (Score:2)
I assume it's "'Free' as in speech, not 'Free' as in beer." He actually spent about $700 on the hardware.
Next headline (Score:5, Insightful)
Next headline: College Student Arrested For Building Autonomous Car That Hit Something
Re: (Score:2)
That's not catchy enough for the Mainstream Media. Trye:
"Berserk Cyborg Car Escapes Lab, Goes on Killing Spree!"
Re: (Score:2)
Re: (Score:2)
Re: (Score:2)
Would it be any different than taping the accelerator to the floor and letting it loose on the road?
It depends on the kind of tape you use, duh.
Re: (Score:2)
Next headline: College Student Arrested For Building Autonomous Car That Hit Something
And the next line: Insurance company refuses to cover damages, clean-up costs, hospital bills, loss of income due to disability and so on. Even if you do eventually win expect to spend a few years in court with a lawyer driving you into bankruptcy first. Also if you're arrested you have the right to a lawyer, not so much in civil court when the insurance company claims you broke the terms, I'm sure they have something in the wall of legalese that will apply.
Re: (Score:2)
And the next line: Insurance company refuses to cover damages, clean-up costs, hospital bills, loss of income due to disability and so on. Even if you do eventually win expect to spend a few years in court with a lawyer driving you into bankruptcy first. Also if you're arrested you have the right to a lawyer, not so much in civil court when the insurance company claims you broke the terms, I'm sure they have something in the wall of legalese that will apply.
I don't know the US laws; in Germany your third party liability has to pay if the car was insured, and the damage wasn't caused intentionally. (And your car is almost always insured, even if you didn't pay your fees; if the insurance company decides to cancel your insurance, they will send a letter to you, and another to the police to seize your car until you insure it again).
That has always covered accidents caused by drunk drivers, by thieves and so on. Because third party liability is a legal requirem
Re: (Score:2)
The story of Geohot's autopilot (Score:5, Insightful)
Geohot, a renowned hacker decided to try making a self driving car. It kinda worked, but it is, well, a hack.
However, when regulators came over and asked him to prove that it was actually safe enough for public roads, he backed down and that's how we got Comma.ai free.
I've nothing against Geohot and Comma.ai, quite the opposite in fact, they are great hackers, in the positive sense. However, when lives are on the line, being clever is not enough, we also need the boring and expensive work to make sure it is safe.
Re: (Score:2)
It's not a terribly difficult problem to get to work 99.5% of the time, but with lives at risk most people aren't too happy with that number. The airline industry has a failure rate of 1 in 10^-13 deaths per passenger mile or something like that.
Re: (Score:2)
The thing is, I doubt it'll ever be 100% safe. There is always a chance of some kind of glitch or bug. If the goal is 100% we can hang it up now. But if you consider how dangerous letting people drive I think I'd be happy with 80%. I see people tweeting and texting all over the place and that's not even talking about people yapping on the phone, eating or reading a damn magazine! Then there are the people that just can't drive. They lack coordination or something. I think 80 percent is better than hu
Re: (Score:2)
Re: (Score:2)
I drive in a car on roads with people that aren't that safe. I'd rather a computer was driving those cars.
Re: (Score:2)
Re: (Score:3)
I notice that many people range from inattentive to reckless. Today I went through a school zone during pick-up time and was driving about 30-35 in a 30 zone, and got my doors blown off by a Yukon that did a quick lane change immediately after passing me due to another "slow" car observing the speed limit. Yes, I know most drivers aren't reckless idiots and it just seems that way but enough are that I believe computer control would reduce accidents and especially severity of those accidents.
Re: (Score:2)
Re: (Score:2)
In 2015 over 35,000 people died in auto accidents and that's doesn't include those that are maimed, some never to walk again. I think autonomous vehicles could bring that number way down.
The difference with robot cars is that the lower number will be purely random.
I'm not sure that sits well with many people.
Re: (Score:2)
In 2015 over 35,000 people died in auto accidents and that's doesn't include those that are maimed, some never to walk again. I think autonomous vehicles could bring that number way down.
The difference with robot cars is that the lower number will be purely random.
I'm not sure that sits well with many people.
Currently the death rate is already largely random - anyone who isn't a driver has virtually no control over the accident rates, and probably a large number of drivers had no real control of the situation once it started going bad. Of course our perception of control might make us feel safer than we would be in situations where we perceived ourselves to be less in control.
That's why we end up with safety legisilation that mandates seatbelts and crash standards and backup cameras that might have little succe
Re: (Score:2)
Currently the death rate is already largely random
No it isn't. There is some random elements, but it you're familiar with crash statistics, you'll know that once you take out the common factors such as alcohol, drugs, mobile phone use, fatigue, bad weather, speeding, faulty vehicle, health issues etc, the chances of death on the road drop dramatically.
Re: (Score:2)
Currently the death rate is already largely random
No it isn't. There is some random elements, but it you're familiar with crash statistics, you'll know that once you take out the common factors such as alcohol, drugs, mobile phone use, fatigue, bad weather, speeding, faulty vehicle, health issues etc, the chances of death on the road drop dramatically.
You missed my point by chopping off what I said. Was that deliberate? It seems needlessly argumentative. The quote continued with "anyone who isn't a driver has virtually no control..."
I don't disagree with the useful information you have added to the discussion.
While you have some control over your own behaviour, your odds of encountering someone else being infleuenced by "the common factors such as alcohol, drugs, mobile phone use, fatigue, bad weather, speeding, faulty vehicle, health issues etc." are la
Re: (Score:2)
You missed my point by chopping off what I said. Was that deliberate? It seems needlessly argumentative. The quote continued with "anyone who isn't a driver has virtually no control..."
They choose who they get in a car with.
We often have an illusion of complete control over our destiny when we are behind the wheel.
This is my point. We don't have complete control, but we have some control. Even as passengers we have a choice in who we drive with. And 'some control' has more emotional value than 'no control'.
Autonomous systems seem likely to reduce many of these common factors in the other cars on the road, so I suspect they will become more common if they in fact do so.
I agree that robot vehicles will have a place, but they will also introduce different problems, so the argument can't be reduced to 'less overall accidents equals win'.
If less overall accidents was the only deciding factor for people's transport choice, then this problem has
Re: (Score:3)
That's deaths per mile not accidents. I remember how people used to die in minor accidents when I was a child. Even a 35mph accident was serious when cars had metal dashes and no seat belts. The hood on cars in the 50s were made of heavy steel and would not bend. In a front end collision it would often break loose and come straight back decapitating the people in the front seat. Now they crumple. In the last few decades it's gotten to the point where most accidents don't even require first aid. Cars a
Re: (Score:2)
Well, sure. But you can inflate the accident rate by many orders of magnitude and still not be close to the parent's 80% number. Whether the number that concerns you is 99.999999% or 99.99999% is basically immaterial; the point is that autonomous navigation has to be very, very, very good to beat human drivers which, crummy as they are, are already astoundingly safe. "One car capable of driving down the street safely many times" is a great technical achievement, but is not the bar being set here.
But the point is we don't really know the safety stats for this type of vehicle until we get a few gizillions of miles of driving. If this type of car performs better than the majority of drivers in the majority of situations, but worse than the majority of drivers in a minority of situations, it MIGHT have worse stats overall on a per-mile basis or it might have better stats overall. While I am not the original poser of the 80% figure, I suspect that what they were aluding to was their perception of the po
Re: (Score:2)
Re: (Score:2)
And we need to compare apples to apples. A human pulls out of their driveway and parks in a dense parking lot. A human drives in all weather. Either you have to take those accidents out of the stats for humans and only apply stats for humans in the conditions where AI can drive, or you need to wait until AI does all these things and see how safe it is.
I don't disagree, but I can certainly imagine that there is a place for a system that can only do "highway driving" and isn't able to do parallel parking (though right now I think we might have the opposite - parking but not driving.)
Re: (Score:2)
One thing for sure, the computers are getting better and the humans are at best about the same as always.
Re: (Score:2)
One thing for sure, the computers are getting better and the humans are at best about the same as always.
Either this human is gettng worse as I age, or I am becoming more aware of how piss-poor I have always been.
Good Enough (Score:2)
It's not a terribly difficult problem to get to work 99.5% of the time
I'd say any technology that is 20% better than humans should just be let out on the roads the way we do 80 year old drivers.
Re: (Score:2)
They weren't even regulations. The government was halds-off the entire thing. They were merely inquiry questions meant to h
Depends on supervision (Score:2)
It's not a terribly difficult problem to get to work 99.5% of the time, but with lives at risk most people aren't too happy with that number.
Depends.
If the system works even 90% of time and there's a human backup that is alert and focused, then it's good already.
(like autopilots found in airplanes, boats, some modern high-speed train.
Autopilots help automating some minute detail of the driving/sailing/flying.
But autopilots are still under the supervision of a human in charge.
It just relieves the human of part of the stupid hard gruntwork.
That's also were Tesla's autopilot and Google's prototypes on highway fell in).
If the system works even 99.9
Re: (Score:2)
Attention (Score:2)
not all humans are capable of staying focused on the ride while not involved in it
Hence some strategies of asking to keep the hands ready on the wheel (and other similar micro-involvements)
(And there is experience, coming from the world of train automation, that suggest that this works (a bit).
e.g.: TGV train operators are required by the system to periodically hold the thrust control wheel)
Also in my personal experience, you still remain involved in the driving :
- even if the adaptive cruise control is taking care of keeping distance with the car in front, you need to periodically adjus
Re: (Score:1)
LoB
The kids are alright (Score:3, Funny)
Young people today are impressive. When I was a senior in college, I was turning milk bottles into bongs.
They were sweet bongs, though.
When I got to grad school, that all changed because I was suddenly surrounded by people smarter than me and I had to actually work. But those first seven years of college were a lot of fun.
Re: (Score:2)
So once you got to grad school you had to make calibrated multichambered bongs with working valves and heat resistant materials?
Re: (Score:2)
Oh, fuck off. Do you have any idea how many technological and scientific advancements were made over the past few decades, when we were told there was, "too much regulation"?
Surprising (Score:3)
Re: (Score:3)
Nope, CANBUS is an open protocol, if you have a real CANBUS reader (not one of those $13 ELM junk) you can read every piece of data coming across the bus. A honda civic is the cheapest car you can buy and when I say cheap this also means they went cheap on security too. Newer e.g. Audi's and Mercedes' have since separated critical components to a secondary bus (which you can access via the ECM under the hood but its no longer as simple as plugging into the ODB2 port).
My question is how is steering done?
Re: (Score:2)
Nope, CANBUS is an open protocol
I like how all the threads keep coming back to pot...
Re: (Score:2)
It only works on Honda Civics with "Honda Sensing", which includes Lane Keeping Assist. So they tap into that for the steering.
So basically you buy a car what already has Adaptive Cruise Control and Lane Keeping Assist. And you buy this box made by an amateur and run open source software on it that claims to be autonomous driving, but is actually just Adaptive Cruise Control and Lane Keeping Assist implemented by an amateur on non-automotive grade hardware.
Pointless as well as dangerous.
Re: Surprising (Score:2)
LKAS just pulses brakes to keep it car in Lane. It detects lanes using a camera.
OpenCV can easily detect lanes, it has been used in many systems in the past. Beyond that, your ABS controller controls braking these days to prevent loss of control and it does this by independent braking of wheels. Similar technology is used to keep your car handling smooth around a turn.
Good to know these guys aren't doing anything unique
Re: (Score:2)
No, both the cars supported use steering for LKAS. That's WHY those are the supported cars.
Re: (Score:2)
if you have a real CANBUS reader (not one of those $13 ELM junk) you can read every piece of data coming across the bus.
Does ELM327 sniffer mode miss packets?
Re: (Score:2)
I imagine that the cheapo bluetooth dongles can't keep up, and probably drop datagrams that are on the bus. A higher-end device could do faster logging.
I suppose that's possible, but I doubt it. Those bluetooth serial devices will usually let you do 230kbps. A $50 OBDLink LX can probably do the job just fine. In any case, I have a USB ELM327...
Re: Surprising (Score:2)
No they cannot. They don't read CANBUS. They just talk OBD-II. CANBUS is a very different protocol. Timing is much different that a cheap Chinese part can't match (yet)
Re: (Score:2)
No they cannot. They don't read CANBUS. They just talk OBD-II. CANBUS is a very different protocol.
OBD-II is not a protocol. It is a connection standard. It implies one of several electrical connection standards, to go with several different protocols. One of the protocols used on OBD-II is CAN, and one chip which supports CAN is ELM327. The ELM327 supports both SAE J2411 (slow, single wire) and ISO 15765-4 (mandatory in all vehicles in the USA since 2008.)
Is your complaint that ELM327 doesn't speak some other protocol commonly being used between modules? That would be unfortunate.
Re: (Score:2)
They have a much improved version on Hackaday right now.
Re: (Score:2)
They have a much improved version on Hackaday right now.
I'm having trouble finding it... My car doesn't even use CAN, it's K-Line, but I'm still interested.
Re: (Score:2)
Sounds like a fascinating project, though I think you were wise to avoid getting yourself into trouble. Interestingly, from TFA:
Bryant Walker Smith, a law professor at the University of South Carolina, says that federal and state laws probably don’t pose much of a barrier to those with a desire to upgrade their vehicle to share driving duties. NHTSA has authority over companies selling vehicles and systems used to modify them, but consumers have significant flexibility in making changes to their own vehicle, says Smith, who advises the U.S. Department of Transportation on law and automation.
However, I think this law professor is still giving bad advice. Simply because the NHTSA won't stop you does not mean that you would have immunity or pity if something went wrong, especially if it hurt other people.
Re: (Score:2)
You have to dismantle part of the dashboard, because they firewall the standard diagnostic port to prevent some random OBD-II dongle spewing bogons that affect steering or something.
The docs for the control electronics are not available without NDA, but you can just sniff the packets off the wire. This is the guy who cracked open the PS3 by physically glitching the memory bus, and no protection scheme ever really works anyway.
Re: (Score:2)
I would have expected fiercely guarded proprietary systems...
They are fiercely guarded by the impenetrable force of security through obscurity. And if there's anything we learnt from the car industry is that this is the only security they have.
3D printed case is the key (Score:2)
I think it's time for PRT (Score:1)
In the 1960s, it was theorized that with a dedicated track, and some computer power, it would be possible for vehicles, holding a few people each, could reduce the need for cars. Various groups around the world never could get all the problems taken care of, including sudden bunching up of traffic.
Now that computing power has exploded, now might be the time to try dedicated PRT roads again. Basically half a ton, unpiloted vehicles, on overhead roads. I am aware it is considered an eyesore, but if robots use
Re: (Score:2)
cpu power will not fix tack switch limits / spacing / slow moving people that can jam up a enter / exit point. rural area covage / transfers from this system to other systems in place. very built up Urban areas with little room for tacks and stations.
The easy part (Score:2)
While what this kid did is impressive, he's only done the easy part: getting a car to drive itself under a limited set of circumstances that he knows about.
The hard part is to get a car to drive itself under all sorts of weather and road conditions, and safely handle all kinds of expected and unexpected road hazards, such as potholes, people, bicycles, and crazy drivers.
Re: (Score:2)
I'd hope the average /.er knows that the hard part to self-driving cars is 'productising' it into something that passes regulations and doesn't kill people. However, I say this guy's pretty cool - first up, he's done some Ultimate Geek Hacking - it involves computers, 3d printing and cars - what's not to love!?. Second, he's shown it's possible for ordinary people to get some distance with self-driving cars. It means Fred-in-his-shed can take a look at this technology and try some stuff out. Sure, most of
Re: (Score:2)
I agree with everything you said, except the last phrase. I think you wouldn't have to be a millionaire to tinker with self-driving technology, but you certainly would need to spend millions to go into business.
Re: (Score:2)
Cars are becoming more automatic anyway (Score:2)
Two days ago I was at a launch party for the 2017 Honda Civic. There was a 1st generation Civic from 1978, my own 1991 4th generation Civic, and several subsequent generation Civics, all in the Honda showroom arrayed around the new model.
What worries me is that so much on the new model has gone automatic. It's got a radar set in the front bumper to measure distance to the vehicle in front. It can be set to automatically speed control itself to maintain safe distance from the vehicle in front. It's got blind
Re: (Score:2)
It was always the common sense option with UK driving tests to take them in a manual gearbox car, so that you'd be qualified for both manual and automatic. But for anyone now who fancies owning an EV, I'd suggest also learning to drive in an EV. Over the next few years EVs are going to take over the entire market. So those skills of clutch control and gear changing and hill starts will probably remain unused.
And quite possibly in 20 years time there will be no point in taking driving lessons at all, because
Example of Government Stifling Innovation (Score:1)
So, they got a letter from big government, and decided to give up and go home? That must have been a seriously threatening letter. How dare they try to innovate and compete against the large corporate oligarchs like Google, Apple, and Tesla?
Re: (Score:2)
this is cool (Score:1)
Uh... (Score:2)
Job security via hobbyist love (Score:2)
He should have no trouble scoring a sweet job...