Excellent write up and project! Wish I had that much patience.
I am suprised it actually worked. Probably required some good connections at a Tesla service facility / input from someone who knows these cars in and out.
Seriously. The thought of this thing being on the road scares the heck out of me. All it takes is for there to be one assumption the autopilot software makes about the hardware that didn't get retrofitted, and you've got an accident on your hands.
People are always so paranoid about autonomous cars, as if we expect cars and drivers to otherwise be infallible, contrary to all available evidence.
People have been modifying their cars since there have been cars. Sometimes the modifications are problematic. Sometimes problems result in accidents. Those cases are rare. Moreover, since the modifications are all bespoke, the odd erroneous one affects only that single vehicle.
On the other hand, the modifications let people learn. It's how we get new ideas. That improve safety. For more than just one car.
> If the modifications a person makes have the potential to kill members of the public, I think you're wrong.
Literally everything has the potential to kill members of the public. A bug in your mobile device app could interact with a bug in the hardware and cause the battery to catch fire in the middle of the night and burn down an apartment building. If you install casters on the office copy machine, it may slide down an incline, through a 10th story window and kill someone in the street.
> This is new technology; laws and regulations need to reach a stage where they're able to protect against the worst from happening.
There are two classic methods of discouraging harm. The original, which works fine for low volume modifications, is that if you do something reckless you get held liable after the fact.
The method you're proposing is to do safety testing ahead of time. But that is only reasonable for products that are mass produced. If you're going to make a million of something, it makes sense to analyze it ahead of time because the high fixed cost of doing that can be amortized over a million units whereas the amount of harm prevented is multiplied by a million units. That method is highly inefficient for one off modifications where the opposite is true.
The complexity of "cars" has gone up exponentially over the years, so modifying them while still being reasonably sure that they're safe is getting harder. Today cars are at least as much software as hardware and it's completely unclear which assumptions are hardcoded into them that you could invalidate by modifications.
> Today cars are at least as much software as hardware and it's completely unclear which assumptions are hardcoded into them that you could invalidate by modifications.
This rings hollow as a novelty. If you replace the transmission with one that has higher gear ratios, it may raise the top speed past the maximum rating of other components. Any modification that makes the vehicle heavier may cause the brakes to be insufficient. Any modification that changes the weight distribution may have a number of difficult to predict consequences for steering and stopping. Tires that are better on a track may be worse in the rain. Tires with a different diameter may cause the speedometer to mislead the driver. Tuning the engine to reach higher RPMs may cause excessive hydraulic pressure and damages the brakes or steering. Anyone making any modification or repair may use non-OEM parts with arbitrarily different specifications.
These are all complicated engineering decisions that depend on the assumptions made by the OEM. Maybe the brakes are sufficient with an extra 700 pounds, maybe not.
Hardware isn't simpler than software, the assumptions are just less explicit. And encoding the complexity in software rather than hardware can make them less error-prone because software is more adaptable. It can read sensors and make realtime calculations. The assumptions encoded in hardware are static.
Hardware is a lot simpler than software, because there are far fewer parts that can interact. Cars run millions of lines of code, they have far fewer mechanical components. There is a reason why software is eating the world, it's because you can have a lot more complexity for a lot less money.
> Hardware is a lot simpler than software, because there are far fewer parts that can interact.
Do you remember the hardware that existed before it was replaced by software? Thick bundles of wires everywhere, mechanical switches operating relays and solenoids, intricate clockwork mechanical parts rotating in synchrony. There were early vehicles with mechanical anti-lock braking systems, cruise control, airbags, etc.
Most of that complexity has been moved into software but the software wasn't what created it.
And before it was software you had the full complexity that comes with the systems not only being designed by humans but also made out of atoms and molecules subject to all the intricacies of physics and chemistry, with no checksums or error correction to detect or correct a failure when something is operating out of spec.
> For over a year, Waymo has been offering trips to the 400-plus members of its Early Rider program who use Waymo’s ride-hailing app to summon the minivans for free trips to school, the mall, the gym, or elsewhere within its suburban Phoenix service area. Soon, Waymo will make that service available to the general public and it will start charging money for it, too.
People drive highly modified cars all the time, and we're not too concerned about it. Certainly I wouldn't trust this autopilot to be foolproof, but that's true of the factory version as well!
Yeah but most people modify their cars in a way that demands more attention while driving and reap the rewards of modification when they're paying attention to the car.
There are some people who modify their cars so driving demands less attention, but hereunto that was never technology that gave software the authority to control how a car cooperates with other cars. The broomstick on the throttle "cruise control" isn't meaningfully comparable to hacked autopilot at the helm.
Sure, if the driver is paying attention there's nothing to worry about. But this system is dangerous because it may lull a driver into a dangerous complacency.
I think you are getting downvoted because you are just wrong - it's not illegal, but you have to tell your insurer about it, as long as they approve it it's absolutely fine to drive. Some modifications(engine swaps for example) require the car to be re-certified by the official authority, but it's not an impossible process, just takes a few weeks and some money.
It is legal if you get it re-certified - you car will most likely receive a different tax band because emissions will go up(or be refused certification if it goes over allowed limits).
I think there's some misunderstanding here - even in the super-strict-and-bad EU you can build a car in your shed and get it registered and drive it. But you can't build/modify your car to be against regulations, be it emissions/noise/pollution or whatever else. As long as it's within the allowed parameters, then knock yourself out - there's a thriving modding scene even in countries with insane regulations like Germany.
That's a non-sequitur. One modification not being legal in your jurisdiction means nothing for other modifications. If you want to get more than a few thousand miles out of a car, you'd better be willing to modify safety-critical parts of it by replacing the brake pads and tyres - plenty of people do that themselves.
The difference is that a highly mechanically modified car is a fairly understandable system.
I write embedded firmware for robots for a living. I've got a fairly good grasp of the complexities of a system like this. These modifications terrify me.
How does he know that one system isn't talking the n-1 protocol which is incompatible in subtle and weird ways from the current one?
How does he know that there isn't calibration data hiding inside one of the systems he installed which assumes a different dynamic response?
Systems engineering for a system of this complexity is really really complicated. Bugs develop in the interactions between systems even when every part is working fine. If you start glueing stuff together which isn't designed to work together and is relying on unpublished unversioned APIs for communication, you're going to get all sorts of problems.
Opening up wire harnesses introduces corrosion - I would not trust an autopilot car with hand soldered wire harnesses. There is a reason Tesla would replace all of the harnesses if they were to do this...Sure its fine now but in a few years those connections will start to fail.
The main reason is that they would value the time taken to modify the harnesses much more highly. As long as the modifications were done well, I wouldn't be concerned. Plus, one would hope that the autopilot system was designed so that in the case of a loose connection it fails in a sane manner.
Disconnecting or cutting the wiring harnesses does not introduce corrosion. Assuming the connections are made well using the correct methods it will not be an issue. From the pictures it looks like they soldered and heat shrunk the connections which should be a long lasting repair.
It absolutely does. A wire inside of a harness is a closed system. You have water tight sealing on the end of the connector on both ends of the cable. As soon as you open the sheath and expose the bare copper to atmosphere, corrosion starts. You cannot splice an automotive harness and keep the same characteristics as when it came from the factory.
None of this is true. Wiring harnesses aren't water tight, primarily to prevent corrosion. If you have a totally water tight harness, and water some how gets in, it has no way to dry and will cause corrosion.
You absolutely can splice a harness and have it have the same characteristics, I do it every day.
We are talking about two different things. Notice I did not say wire harnesses are water tight. Each individual wire inside the harness - the copper inside - is sealed to the environment. You will get corrosion on copper if you expose it to atmosphere - exactly what happens when you splice it. By the way - I do development / testing of automotive electronics and am well aware of the specs / test procedure and design decisions of wire harnesses...
Most automotive wiring harnesses have crimped connectors, these leave the ends of the wires open to the atmosphere. Connectors used under the hood tend to have seals these effectively seal the wires as well. Wiring used inside the cabin (or anywhere not exposed to the elements) generally don't have any seals to save cost. As long as you're alright at soldering and heat-shrink the wires they'll be fine, this really isn't much of a problem.
Hand soldered wire harnesses are still used heavily in the industry. If you do not have the right crimper, J hooks and direct soldering are probably even more reliable than crimps (hence why official crimpers are so stupidly expensive). I think IPC has some standard about hand splicing. That being said, I would of probably used the heat shrink solder combos.
Sort of agree, sort of not. There are people who can do such modifications safely. The problem is, one might have the belief of belonging to said group, while being mistaken.
I'd say the problem is that it is pretty much impossible to make such an assumption in the first place without extensive inside information.
I would not be surprised if there is not one single person at Tesla that would be confident in knowing each possible pitfall from doing this. And I would certainly not expect anyone from the outside spending a few weeks on this to gain that knowledge either.
Very impressive, but in my eyes he have ruined a perfectly fine Tesla.
"Your scientists were so preoccupied with whether or not they could, they didn’t stop to think if they should."
As others have noted, this seems like a terrible idea. One small mistake, and it could kill his wife, himself, or someone else on the road. All the other self driving Teslas have at least gone through testing to make sure they are road safe.
I know I'll be avoiding any yellow Teslas from now on!
NO it's way worse than driving in general, because once the car has been modified, its behaviour on the road and that of the guy in the driver seat are completely decorrelated.
Ergo, the guy can be much more alert than most people on the road and STILL crash into others without any sort of warning, simply due to a software bug.
For various reasons, in regular cars, this should not happen.
If they keep their hands on the wheel and the self driving actuators can’t overpower standard human strength, I would be fine with it on the road near me.
It also looks like his workmanship was reasonable so he probably did everything correctly.
> This required transplanting the controls from a salvage steering wheel, which had a blown airbag, into my wife’s steering wheel.
Reusing delicate parts from a salvage wheel that had a blown airbag does not seem particularly safe to me. Then again, at least it only endangers his presumably consenting wife and not third party drivers/pedestrians on the road.
> Reusing delicate parts from a salvage wheel that had a blown airbag does not seem particularly safe to me.
Why not? The one part that matters for safety in a wheel is not delicate at all: it's the connection to the axle. Even if all power is suddenly dead, the vehicle can still be manually controlled.
As for the airbag, I guess (and hope) that the author did replace it.
>Reusing delicate parts from a salvage wheel that had a blown airbag does not seem particularly safe to me.
Well the existence of the entire used car parts industry seems to indicate otherwise. For some parts, especially out of production control modules over various kinds, used is the only option. Mechanical failure is a drop in the bucket in terms of road safety. Mechanical failure as a result of substandard parts (as opposed to deferred maintenance/repair) is such a small fraction of that drop I've never heard of anyone even trying to measure it specifically.
If... The whole premise of tesla lane guidance is to allow for the driver to relax. This allows for the driver to not pay attention which in turns has caused several deaths.
That is completely inevitable. Humans are just not capable of behaving rationally in such a system.
These modifications increase that risk. These modifications do not increase the attentiveness of the driver. Step 3 ... Step 4: profit?
Let me describe to you the 900-kilogram car I'm building, with forged pistons and a turbo big enough to fit your head inside. It probably won't hold traction until third or fourth gear. One small mistake and it would kill all sorts of stakeholders.
People like me drive highly modified cars day in, day out. Though I'd honestly trust the handiwork of someone who can figure out the sophistication behind a tesla and it's electronics more readily than I'd trust your general backyard mechanic.
It's not uncommon to see 40 year old cars driving around with stock suspension and brakes, that have had a 6.2L LSx engine fitted (as an example). I'd be more concerned with those kinds of conversions before I'd worry about the one Tesla in the united states that has a jerry-rigged autopilot.
> Let me describe to you the 900-kilogram car I'm building, with forged pistons and a turbo big enough to fit your head inside. It probably won't hold traction until third or fourth gear.
Sarcasm is hard to pick up via a written media so i'm going to assume you're legit. Where is this thing even legal to drive??
What he described sounds like a lot of street legal cars in the US. As long as it has turn signals, windshield wipers, seat belts, and some other things on a short list, it's allowed on the road.
Yeah not to mention guys like Saleen and Hennessey who make some downright ludicrous road cars. When you're talking about the power levels they make, the difference between 900kg and 1200kg is negligible :P
Edit, and a more serious note, this is the flip side of the kinds of things we're discussing though. Not all modified cars are built to equal safety considerations, and not all modified car stories end well.
Racetracks, mainly. While I'm keeping the car registered, I'm sure the first time I have a problem with a police officer the car will receive a defect notice and won't be legal on the road any longer until it passes an engineering inspection. I'd need to be doing something pretty silly to get pulled over though, as it's all in a mx5 (miata) and it's a bit of a "sleeper" - i.e. it doesn't appear to be modified when looking from the outside.
Interestingly enough though, the work I'm doing should be able to pass an engineer if required, but that's an investment I'll make when the defect arises. In Australia it's pretty commonplace to daily drive your race car until it gets defected, haha.
It should be noted though, that in certain states in the United States, it appears (to an outsider) that some extremely powerful drag cars etc can be driven on the streets. Or, it's a similar scenario, in that they shouldn't be driven on the streets but are anyway...
What doesn't make any sense to me though that even if you avoid the police and any inspections, doing this to a car would 100% invalidate any insurance you have on it - making it an extraordinarily stupid idea to drive on any sort of public road, because if you get into an accident any insurance company will just refuse to pay out a single penny since the vehicle has been heavily modified past the factory specs - unless you declare all modifications to your insurer, which I am assuming you don't, since you also mention avoiding a full engineering inspection.
I'm with Shannons and have all modifications declared. The engineering side of things is primarily for standard road/vehicle registration and checks that you're compliant with ADR (Australian Design Rules). i.e. my rollcage is CAMS (Confederation of Australian Motor Sport) compliant, however for it to be ADR compliant I would need it inspected by an engineer, while Shannons still happily covers it.
But yeah my insurer knows everything right down to the brand of main bearings I bought.
All they need to do is prove that any of the modifications made the car unsafe to drive, or that it wouldn't have passed the technical inspection with them in - they are willing to cover mods, but it always comes with a clause of "as long as the car remains legal to drive on the road"(which they don't check - they just acknowledge the existence of modifications, nothing else).
Also parts of Europe if you're willing to jump through hoops, probably also street legal in less developed nations that don't regulate that sort of thing.
I drove a heavily modded car when I was 17, so I'm familiar. We bored out the pistons to make a 2.8L engine closer to a 4.2L and then dropped the cat and replaced the 4 speed gearbox with a five speed manual.
But it was done by an experienced mechanic, and pretty much every failure mode resulted in harm to the driver, not to others.
This guy is a hobbyist and he's mucking with hardware that is controlled by software, which will magnify any problem he may have created.
Manual mods to a car are wholly different than modding hardware that supports a complex software system that he can't modify or even read.
Most of the time, yes. Its advantage was accelerating off the line, a capability which I rarely used.
I see what you're getting at though, that the modifications made it less safe on the road. I don't disagree. It was unsafe. I shouldn't have been allowed to drive it.
The difference here to me is the more normal crazy car modification scene isn't mucking around with systems that take control of the car, it's modifications to how the car operates and handles but it's the 'driver replacement' that Tesla's autopilot is used as (as much as the companies try to insist that it's an assist we've seen people don't pay as much attention to these semi-autonomous cars as they should as the driver). So we're talking about a system that already has problems with errors and driver attention being patched together where the myriad subtle errors will be completely hidden from the user.
That's the fundamental difference to me between this kind of mod and the more traditional soup up etc modifications.
I'd only really call the brakes there safety-critical but either way they're at the base layer pretty simple hydraulic systems that are very well understood (and there's always the e-brake which doesn't usually get screwed with). Even that doesn't get around the fact that these systems even in their full factory installed working order have driven directly into the sides of semi trucks, lane dividers and into the back of completely stationary vehicles.
Look at the Toyota unintended acceleration incident to see how dangerous a malfunctioning engine can be. (Even if it ended up just being floor mats or driver error, it shows what can happen in that case.)
Other cars in full working order drive into stationary vehicles and lane dividers every day. Nobody’s freaking out about DIY mods for those.
People give "experienced mechanics" far too much credit. Sure, many of them are great and have fantastic diagnostic skills, but many of them have also been fixing the same 10 problems for 20 years.
Case in point: years ago I took my (modified) pickup truck to a transmission shop because it was leaking gear oil. I had the Factory Service Manual and knew exactly what the problem was, but didn't have the tools or interest in doing this particular repair (required dropping the transmission and I knew from experience what a pain in the ass it was without the right tools).
The transmission shop refused to believe me when I told them what the problem was: "in all my years, I've never seen that happen" etc. They even made me sign a document to guarantee that I'd pay them after they pulled the transmission and I turned out to be wrong.
Result? The shop manager came out shaking his head, "sonofabitch, you were right. I have never seen that before. It's an easy fix, but we need to keep the truck overnight so we can order the part."
There are many, many cases of an untrained individual whose self-interest leads them to research and become deeply educated on something important to them that the so-called experts have only a passing familiarity with. Usually it's something medical-related, in this case it was automotive.
For any accidents I can think of where this modification would cause an accident, the driver would have to sit back and let the autopilot do the driving: that's irresponsible in any car. Is there something else, like severing a drive-by-wire connection to the brakes, that I'm missing?
He has his own fancier Tesla, you would think that they would swap so she has the better car. The yellow colour is just a wrap, he could have had things cosmetically correct for her by wrapping his P95(D). Plus she would not have been off the road waiting for the homebrew modifications to happen.
I am okay with people modifying their cars, in fact I like to go to the race tracks where people go at actually ludicrous speeds in their modified cars, repairing them between races. But it is a different thing modifying a car for someone else in an experimental way. On a track where the driver is a 'racing driver' is one thing but, on the highway when the driver is a 'civilian', well, that is something else.
A small rattle means something to the guy who built it and knows what that rattle can be from, another 'civilian' driver won't necessarily know the problem space.
That “one small mistake” bit applies to driving almost any car in almost any circumstances. And yet we allow complete morons to drive them after passing a cursory test.
My ACC/lane assist would do the same thing. Autopilot was never intended to be used without looking at the road. People have misconceptions about the word "autopilot" - it has never ever meant full or even partial autonomy, plane autopilots are way less autonomous than a basic ACC in a car is.
Citation needed. In the history of lane assist, Teslas are the only cars that have actually run into center dividers and killed their drivers while lane assist was engaged, because other automakers are sensible enough to disable lane assist where the lane markings are ambiguous.
I drove some 2-door Volvo that almost sent me into a truck that was next to me in the middle of a turn on a highway. Actually it didn't work in any turns (as in did nothing, most of the time). Terrible thing, especially considering that there was absolutely no clear mark anywhere on the dashboard that it's enabled and I was driving 200 km/h (on autobahn in Germany) when the situation with a truck happened. I also never found out how to disable that thing, thankfully I didn't have to drive on the way home.
I own a Ford Mondeo (the 2008 high end variant) and it would not disengage when marks on the right side disappear, left is enough - but that is a problem because it'll resist if I try to go to the right (if there is a split) and I guess that a very similar accident to the Tesla one could happen but we have no places like that around here so I can't test. I don't use the system.
Amazingly detailed write up. I would love to know the reference material used to figure out the pinout's of the control units. One of the things I miss from older cars is how hackable they were. It feels like new cars are just a bunch of black boxes strapped together with no labels on them, so reading someone able to actually mod one is super impressive.
> First, don’t expect this to open the floodgates for people to start retrofitting autopilot on to older Model S. This was a pretty crazy project to take on, and honestly I’m not going to do it again. It requires the ability to modify the vehicle’s internal configuration and the ability to calibrate and VIN-burn various modules once they’re installed and wired, none of which I am disclosing methods for accomplishing, sorry.
:/
The difference between showing off and publishing research, I guess.
Yeah, its not great, but unsurprising considering how this person's website has a floating bottom bar asserting that all rights are reserved. Just looking to show off, not help others follow in their footsteps.
Well, VIN-burning could be used to launder stolen or un-QA'd parts from salvage-titled vehicles, so it's understandable to refrain from publishing the methods used.
Jailbreaking can be used for piracy. Exploit code can be used for illegal access. Lock picks can be used for burglary. Guns can be used for murder. Cars can be used to rob banks. Radios can be used to jam communications. Encryption can be used for child porn.
We’d better not discuss methods. Hackers might pick them up.
Auto-Driving cars have the potential to help the world cut carbon emissions. Cars that can synchronise their travel, and coordinate their movements can use fuel more efficiently.
People are not good at driving efficiently. Over-braking, over-idle, under-steer .. all of these things contribute to fuel waste at scale.
The thing that's most compelling to me is the ability to accelerate slowly and maintain a slower speed.
I can't do that. Well, I can, but I don't. When I'm driving, I like to get up to speed fairly quickly, and I tend to drive at or just above the posted limits in most cases. It's a flaw in my character.
But in a self-driving car, not only am I not in charge of those decisions, I'm okay with it. Because I'm not preoccupied with the road and traffic, it's an out-of-sight-out-of-mind situation as I watch my movie or edit my spreadsheet.
Just like it doesn't bother me when the train or bus is slow to get up to speed.
That sounds likely to be massively cancelled out by people taking car journeys that otherwise wouldn't happen because they don't have to be awake, sober or focused on driving for hours at a time, because the cheapest autoparking is way outside the city centre or because the taxi/courier is really cheap now the driver salary isn't included.
And electric vehicles are already relatively efficient in energy use even if driven badly, particularly if charged using renewables.
Could also go the other way: people ride-share more frequently, even as a socialising activity, since they can sit and talk with each other while the car does all the work ..
And, the next obvious step: private car ownership becomes a distant memory.
People don't need alcohol to be unable to drive, or be in a state where they shouldn't be driving. After a hard day of work, people are often too tired to drive; Elderly people often don't feel comfortable driving (often still do because it's their only way of getting around); People with disabilities often can't drive, or need expensive adaptions to be able to. Children can't drive (legally) at all.
These are just some of the use cases. But also people who are fully capable of driving get advantages, particularly regarding commuting - instead of sitting in the car for an hour, one can rest, or if the job allows it, work on the way to the office.
Additionally: we, as humans, suck at driving. And whilst it's a valid question whether machines can currently do a better job at it, I'm glad that there's time and money spent towards a future without human drivers (apart from driving as recreational activity).
Imagine how beneficial full self-driving will be for people too old to drive, which, in an America where automobiles are necessary for transport in many places, causes them to lose their independence. Same for the disabled.
I wouldn't buy it. It's been modified in a way that makes it very difficult for someone to find a person willing and capable of doing maintenance on it. If something in a system he retrofitted breaks, he will just about be the only person capable of fixing it.
When you leave the envelope of normal on a vehicle, it rarely increases the value.
I was not expecting the massively negative responses here. For a site called “Hacker News,” you folks sure seem to stand in complete opposition to any hacking that’s outside your comfort zone.
People are afraid of stuff they don't know about. Most of us work in offices and don't even own a 3/8 socket set. (Of course they have no problem with some dude who barely reads English, or whatever language the service literature is in, performing service on their vehicle because he's a "professional"). There's also the fact that with physical objects you can't just revert them if you do it wrong like you can with code. People here are used to having that capability. Working with things where you don't have a near-zero cost option to revert is going to make a lot of people here uneasy. There's a vocal minority here who just hates cars to begin with and has no problem saying things that feed those fears. Then there's the fact that lot of people here are afraid of liability so they start going crazy when any entity that isn't a multi-million dollar company they can sue starts modifying anything tangentially involving safety even if the risks involved are on the order of being eaten by a shark on dry land in Kansas.
Tl;DR the subject matter and the user-base of HN react in a way that brings out more negativity than each one could normally on its own.
This isn’t my story, I’m just an early upvoter hoping to come back to find interesting commentary instead of a bunch of whining.
I think your assessment is correct. Adding even more negativity to the mix, a lot of HN users are skeptical of Tesla in general and Autopilot in particular.
Autopilot continues to be advertised [0] as if it were fully or even near-autonomous driving, so I think those of us who know better are reasonable to start from a position of skepticism.
Gives a whole new meaning to 'move fast and break things'.
But seriously I find the backlash so interesting and I think it is mainly for the reasons you describe. We all know how bug-laden much of the software that's shipped is, but it doesn't really matter because (for the most part) it's easy to iterate, issue an update - to the point that people who run un-updated software are considered foolish, etc etc. I think we should remember that actual _engineers_ are out there who design and make the bridges, cars and buildings that keep us alive who don't have this luxury. This guy seems like one of them and is clearly skilled (you can see that his workmanship looks good and he's taking care in these photos) and should be respected in my opinion.
I'm not sure I understand why someones English reading comprehension should matter when it comes to servicing a vehicle if they are an experienced mechanic at a reputable shop.
Because that's the language that service literature in the US typically available in.
In my opinion literacy at all is optional for 99.9% of automotive work but the kind of people who will pearl clutch over an enthusiast rebuilding calipers in their garage will probably not be all that pleased to learn that it's not uncommon for someone who can barely read the service literature to be servicing vehicles and I thought it was worth mentioning.
I think you'll find similar pushback to any form of hacking that endangers other people's lives.
If we were talking about removing the safety valves on a steam locomotive so it could run hotter and faster you would see similar pushback from people who know what boiler explosions can do.
There are stories on here from time to time about homemade radio systems of various kinds. The risk to life (interfering with aircraft or police radios, for example) sometimes gets mentioned but is never the focus.
Maybe it has something to do with grafting a lane-keeping cruise control system onto a multi-ton vehicle with lithium onboard, and testing on the open road being a Bad Idea™?
If he was testing at the track, at least he's only risking people that he knows.
If it gets something as small as a fender bender, and his insurance gets a whiff of this, you'll bet that they'll try hard to null-and-void the coverage.
Millions of people graft ad hoc, poorly designed lane keeping and cruise control systems into their cars every day. The result is tens of thousands of fatalities each year in the US alone. Yet we seem to be ok with that.
I think this statement requires a source, millions of people are hacking auto-pilot like features each day, causing tens of thousands of fatalities? I don't know if I'd be ok with that, but I don't think it's true.
I don't get it either - the actual risk of this guy testing on the open road is pretty minimal, he could resume control of the car at any point should things have gone sideways.
Zero risk often achieves zero results - you have to balance your risk with other factors.
Great write-up. Elon talked last week about Tesla cars constantly evolving. Sounds like he figured out workable combinations of different parts from the software update metadata. Nifty.
Also expecting it to be blacklisted by Tesla next week.
Sometimes even well publicized articles don't reach just the right person for action to take place. On top of that company culture and their legal stance changes over time, they could have seen it in 2016 and figured it's not a big enough issue to their liability to ban the guy that did this.
Okay so provide us some evidence that this time it reached right person or whos working at their legal department now that didnt before that you assume this time it will get banned.
You misunderstand I'm not saying it has all I was saying was the fact that it was from 2016 and he hasn't been banned doesn't me he won't be. Company's tolerance for things like this wax and wane over time and sometimes news flies under their radar.
So many people here shitting on this guy but it's going to be the future pretty much. Hell I have a GM truck and to add a front camera (for off roading) I'd have to pay like $2,000 for a 480p camera...
That said this guy is brave, not for the mod, but for posting pretty much all of the vin, the name of his car, and showing the color. Will not be hard for tesla to find it and blacklist charging, updates, parts, etc. Musk's ban hammer will be swift and harsh.
The ban hammer would be my concern posting this blog. He would probably get away with it if he didn't publicize it, but my understanding is Tesla has no problem banning cars that it deems unauthorized from receiving updates, using supercharging, etc.
'The idea of buying a car and being afraid to modify it because of how the manufacturer might react feels entirely dystopian to me'
That is exactly where we have been going. for years if you remove the plastic engine cover on Porsches you risk voiding your rights. Now we have the capability of cars that phone home with unauthorized mods being made illegal at the DMV, made illegible for sw updates etc etc. This is another area where our digital rights to privacy and ownership are under serious assault, much like smart phone OS bricking. You own it but you don't have any rights.
A friend of mine is highly regarded engineer at one of electric car manufacturer. He supposed to be getting 8% corp discount for buying their vehicle. He doesnt so I asked him why since he’s so lucky. He told me eventually the day will come that local government or LE will force them to update their car’s OS to report continuesly on the speed you drive, similar to how tachometers work on trucks where cops stop you and can review your speed for the last 14 days you drove your rig nd either report you to your employee or guve you a retro-ticket for speeding. This update, he says will come and nothing you can do to stop it. I take my friend’s word very seriously although getting EV is as tempting as never before!
They already do. Most modern cars have what's effectively a black box data recorder in them. Unlike Tesla however, most cars cannot receive over the air updates and require you to take them to the dealership.
Maybe I’m wrong, but passing laws to centrally limit top speeds in the car loving United States is probably only marginally easier for me to imagine passing than meaningful gun legislation.
I think the car recording the historical data and the police using this as evidence is the much more likely future scenario.
Let’s not forget speeding fines are a revenue generator...
As for life saving, maybe, but I’d need to see a lot of evidence. Modern cars are vastly easier to drive and safer at speed than cars 40 years ago, but we still often keep largely the same speed limits. 155mph in your modern German luxury car practically feels serene.
This might turn out to be true... But today we are clearly in a "golden age" of performance cars. If you are interested in this kind of thing, make hay while the sun shines!
Surprising rhetoric in the comments. Serious question, at what point does working on a car become irresponsible? Does changing the brakes or tires count? How about changing the power steering fluid? Both of these examples can kill yourself or others if performed incorrectly. Should only mechanics be allowed to change this then? I wonder what is the threshold for "common sense" modifications and what is "an easy way to kill him and his wife"?
Should people be allowed to build their own computer or upgrade/install software? /s Working on cars is no harder than working on other things especially when they are just swapping components.
If you build a computer and it fails, I doubt most people will care.
If you try to fix a car and fail, you could end someone's life.
While I'm not against these kind of projects, to equate these things is ridiculous. Where is the line drawn? Vehicles are far more dangerous than your own PC.
I'm edging on hyperbole here but wanted to play Devil's Advocate. You can easily misconfigure a computer build and have it be part of a botnet that does substantial harm and potentially causing death. In fact, there are millions of people who have such computers.
Vehicles are dangerous to the few people around you, but a PC can be dangerous to millions.
At least in the case of an unsecured computer, there needs to be a malicious actor added who will take control and direct the botnet. But in the case of a car, the negligence of the owner can lead directly to fatality without anyone involved being malicious.
This is amazing! Takes serious guts to open up a car like that! Of course there's lots of questions like how did you recalibrate the radars later on? And also how did you get hold of that new firmware to flash?
I love it! the last time I tinkered with cars, I was boosting and racing sufficiently powerful 4 bangers. Glad to see that hacking and modding electric cars is going to be a thing and the future generation can keep the tradition going.
'Boosting' and racing cars within their designed safety parameters and to defined racing scrutineering approval is one thing, grafting very immature self driving technology into an early version of a mass produced EV and then driving it on public roads is something very different. I say this as a very current car builder and modifier.
I put different, more powerful engines in street cars too, also update drive trains and brakes. My point is there is a world of difference between generally understood car building safety metrics and what the insurers are ok with and grafting immature 'self driving' tech into a vehicle.
I'm not knocking it or saying it's wrong, just a very different project with very different safety implications
Can't see the point on publicizing this. There is so much that can go wrong; auto steering could fail due to a slight misalignment in sensors, noise from the custom wiring, any kind of mismatch in the hardware or software. It is irresponsible to use the feature on public roads.
Tesla has _all_ the reasons in the world to find this car and take it out of service. On top of that, he killed the resale value.
166 comments
[ 4.1 ms ] story [ 158 ms ] threadI am suprised it actually worked. Probably required some good connections at a Tesla service facility / input from someone who knows these cars in and out.
People have been modifying their cars since there have been cars. Sometimes the modifications are problematic. Sometimes problems result in accidents. Those cases are rare. Moreover, since the modifications are all bespoke, the odd erroneous one affects only that single vehicle.
On the other hand, the modifications let people learn. It's how we get new ideas. That improve safety. For more than just one car.
This is new technology; laws and regulations need to reach a stage where they're able to protect against the worst from happening.
People have to take driving tests. Is there an equivalent safeguard for an autonomous vehicle?
Literally everything has the potential to kill members of the public. A bug in your mobile device app could interact with a bug in the hardware and cause the battery to catch fire in the middle of the night and burn down an apartment building. If you install casters on the office copy machine, it may slide down an incline, through a 10th story window and kill someone in the street.
> This is new technology; laws and regulations need to reach a stage where they're able to protect against the worst from happening.
There are two classic methods of discouraging harm. The original, which works fine for low volume modifications, is that if you do something reckless you get held liable after the fact.
The method you're proposing is to do safety testing ahead of time. But that is only reasonable for products that are mass produced. If you're going to make a million of something, it makes sense to analyze it ahead of time because the high fixed cost of doing that can be amortized over a million units whereas the amount of harm prevented is multiplied by a million units. That method is highly inefficient for one off modifications where the opposite is true.
This rings hollow as a novelty. If you replace the transmission with one that has higher gear ratios, it may raise the top speed past the maximum rating of other components. Any modification that makes the vehicle heavier may cause the brakes to be insufficient. Any modification that changes the weight distribution may have a number of difficult to predict consequences for steering and stopping. Tires that are better on a track may be worse in the rain. Tires with a different diameter may cause the speedometer to mislead the driver. Tuning the engine to reach higher RPMs may cause excessive hydraulic pressure and damages the brakes or steering. Anyone making any modification or repair may use non-OEM parts with arbitrarily different specifications.
These are all complicated engineering decisions that depend on the assumptions made by the OEM. Maybe the brakes are sufficient with an extra 700 pounds, maybe not.
Hardware isn't simpler than software, the assumptions are just less explicit. And encoding the complexity in software rather than hardware can make them less error-prone because software is more adaptable. It can read sensors and make realtime calculations. The assumptions encoded in hardware are static.
Do you remember the hardware that existed before it was replaced by software? Thick bundles of wires everywhere, mechanical switches operating relays and solenoids, intricate clockwork mechanical parts rotating in synchrony. There were early vehicles with mechanical anti-lock braking systems, cruise control, airbags, etc.
Most of that complexity has been moved into software but the software wasn't what created it.
And before it was software you had the full complexity that comes with the systems not only being designed by humans but also made out of atoms and molecules subject to all the intricacies of physics and chemistry, with no checksums or error correction to detect or correct a failure when something is operating out of spec.
> For over a year, Waymo has been offering trips to the 400-plus members of its Early Rider program who use Waymo’s ride-hailing app to summon the minivans for free trips to school, the mall, the gym, or elsewhere within its suburban Phoenix service area. Soon, Waymo will make that service available to the general public and it will start charging money for it, too.
There are some people who modify their cars so driving demands less attention, but hereunto that was never technology that gave software the authority to control how a car cooperates with other cars. The broomstick on the throttle "cruise control" isn't meaningfully comparable to hacked autopilot at the helm.
Sure, if the driver is paying attention there's nothing to worry about. But this system is dangerous because it may lull a driver into a dangerous complacency.
By doing so you are putting out toxic chemicals because you removed the muffler, this is not legal in states where an emissions test is done.
I think there's some misunderstanding here - even in the super-strict-and-bad EU you can build a car in your shed and get it registered and drive it. But you can't build/modify your car to be against regulations, be it emissions/noise/pollution or whatever else. As long as it's within the allowed parameters, then knock yourself out - there's a thriving modding scene even in countries with insane regulations like Germany.
I write embedded firmware for robots for a living. I've got a fairly good grasp of the complexities of a system like this. These modifications terrify me.
How does he know that one system isn't talking the n-1 protocol which is incompatible in subtle and weird ways from the current one? How does he know that there isn't calibration data hiding inside one of the systems he installed which assumes a different dynamic response?
Systems engineering for a system of this complexity is really really complicated. Bugs develop in the interactions between systems even when every part is working fine. If you start glueing stuff together which isn't designed to work together and is relying on unpublished unversioned APIs for communication, you're going to get all sorts of problems.
I honestly wouldn't be surprised if that was not the case. Tesla is not exactly known for its rigorous approach to programming.
I would not be surprised if there is not one single person at Tesla that would be confident in knowing each possible pitfall from doing this. And I would certainly not expect anyone from the outside spending a few weeks on this to gain that knowledge either.
Very impressive, but in my eyes he have ruined a perfectly fine Tesla.
As others have noted, this seems like a terrible idea. One small mistake, and it could kill his wife, himself, or someone else on the road. All the other self driving Teslas have at least gone through testing to make sure they are road safe.
I know I'll be avoiding any yellow Teslas from now on!
I believe you have just described driving in general.
Ergo, the guy can be much more alert than most people on the road and STILL crash into others without any sort of warning, simply due to a software bug.
For various reasons, in regular cars, this should not happen.
It also looks like his workmanship was reasonable so he probably did everything correctly.
Reusing delicate parts from a salvage wheel that had a blown airbag does not seem particularly safe to me. Then again, at least it only endangers his presumably consenting wife and not third party drivers/pedestrians on the road.
Why not? The one part that matters for safety in a wheel is not delicate at all: it's the connection to the axle. Even if all power is suddenly dead, the vehicle can still be manually controlled.
As for the airbag, I guess (and hope) that the author did replace it.
Well the existence of the entire used car parts industry seems to indicate otherwise. For some parts, especially out of production control modules over various kinds, used is the only option. Mechanical failure is a drop in the bucket in terms of road safety. Mechanical failure as a result of substandard parts (as opposed to deferred maintenance/repair) is such a small fraction of that drop I've never heard of anyone even trying to measure it specifically.
That is completely inevitable. Humans are just not capable of behaving rationally in such a system.
These modifications increase that risk. These modifications do not increase the attentiveness of the driver. Step 3 ... Step 4: profit?
People like me drive highly modified cars day in, day out. Though I'd honestly trust the handiwork of someone who can figure out the sophistication behind a tesla and it's electronics more readily than I'd trust your general backyard mechanic.
It's not uncommon to see 40 year old cars driving around with stock suspension and brakes, that have had a 6.2L LSx engine fitted (as an example). I'd be more concerned with those kinds of conversions before I'd worry about the one Tesla in the united states that has a jerry-rigged autopilot.
Sarcasm is hard to pick up via a written media so i'm going to assume you're legit. Where is this thing even legal to drive??
Here's one example: https://en.wikipedia.org/wiki/Ariel_Atom
EDIT: And another: https://en.wikipedia.org/wiki/Mazda_MX-5_(NA)
https://en.wikipedia.org/wiki/Hennessey_Venom_GT
Edit, and a more serious note, this is the flip side of the kinds of things we're discussing though. Not all modified cars are built to equal safety considerations, and not all modified car stories end well.
https://jalopnik.com/man-dies-after-crash-in-home-built-hond...
Racetracks, mainly. While I'm keeping the car registered, I'm sure the first time I have a problem with a police officer the car will receive a defect notice and won't be legal on the road any longer until it passes an engineering inspection. I'd need to be doing something pretty silly to get pulled over though, as it's all in a mx5 (miata) and it's a bit of a "sleeper" - i.e. it doesn't appear to be modified when looking from the outside.
Interestingly enough though, the work I'm doing should be able to pass an engineer if required, but that's an investment I'll make when the defect arises. In Australia it's pretty commonplace to daily drive your race car until it gets defected, haha.
It should be noted though, that in certain states in the United States, it appears (to an outsider) that some extremely powerful drag cars etc can be driven on the streets. Or, it's a similar scenario, in that they shouldn't be driven on the streets but are anyway...
But yeah my insurer knows everything right down to the brand of main bearings I bought.
They'll happily take your money, the true test is whether they'll pay up if you have a significant claim (e.g. totaled a Bentley :) )
'merica!
Also parts of Europe if you're willing to jump through hoops, probably also street legal in less developed nations that don't regulate that sort of thing.
But it was done by an experienced mechanic, and pretty much every failure mode resulted in harm to the driver, not to others.
This guy is a hobbyist and he's mucking with hardware that is controlled by software, which will magnify any problem he may have created.
Manual mods to a car are wholly different than modding hardware that supports a complex software system that he can't modify or even read.
I see what you're getting at though, that the modifications made it less safe on the road. I don't disagree. It was unsafe. I shouldn't have been allowed to drive it.
That's the fundamental difference to me between this kind of mod and the more traditional soup up etc modifications.
Other cars in full working order drive into stationary vehicles and lane dividers every day. Nobody’s freaking out about DIY mods for those.
Case in point: years ago I took my (modified) pickup truck to a transmission shop because it was leaking gear oil. I had the Factory Service Manual and knew exactly what the problem was, but didn't have the tools or interest in doing this particular repair (required dropping the transmission and I knew from experience what a pain in the ass it was without the right tools).
The transmission shop refused to believe me when I told them what the problem was: "in all my years, I've never seen that happen" etc. They even made me sign a document to guarantee that I'd pay them after they pulled the transmission and I turned out to be wrong.
Result? The shop manager came out shaking his head, "sonofabitch, you were right. I have never seen that before. It's an easy fix, but we need to keep the truck overnight so we can order the part."
There are many, many cases of an untrained individual whose self-interest leads them to research and become deeply educated on something important to them that the so-called experts have only a passing familiarity with. Usually it's something medical-related, in this case it was automotive.
He has his own fancier Tesla, you would think that they would swap so she has the better car. The yellow colour is just a wrap, he could have had things cosmetically correct for her by wrapping his P95(D). Plus she would not have been off the road waiting for the homebrew modifications to happen.
I am okay with people modifying their cars, in fact I like to go to the race tracks where people go at actually ludicrous speeds in their modified cars, repairing them between races. But it is a different thing modifying a car for someone else in an experimental way. On a track where the driver is a 'racing driver' is one thing but, on the highway when the driver is a 'civilian', well, that is something else.
A small rattle means something to the guy who built it and knows what that rattle can be from, another 'civilian' driver won't necessarily know the problem space.
Like the one that ran directly into a lane divider at highway speed? Or the one that ran into the side of a semi trailer?
Citation needed. In the history of lane assist, Teslas are the only cars that have actually run into center dividers and killed their drivers while lane assist was engaged, because other automakers are sensible enough to disable lane assist where the lane markings are ambiguous.
I own a Ford Mondeo (the 2008 high end variant) and it would not disengage when marks on the right side disappear, left is enough - but that is a problem because it'll resist if I try to go to the right (if there is a split) and I guess that a very similar accident to the Tesla one could happen but we have no places like that around here so I can't test. I don't use the system.
:/
The difference between showing off and publishing research, I guess.
We’d better not discuss methods. Hackers might pick them up.
Some people might feel fine with the knowledge that their work might be used for evil, some not.
Despite the billions of $ being poured into autonymous vehicles I've yet to find anyone who actually sees the need, or even less so would buy one.
People are not good at driving efficiently. Over-braking, over-idle, under-steer .. all of these things contribute to fuel waste at scale.
I can't do that. Well, I can, but I don't. When I'm driving, I like to get up to speed fairly quickly, and I tend to drive at or just above the posted limits in most cases. It's a flaw in my character.
But in a self-driving car, not only am I not in charge of those decisions, I'm okay with it. Because I'm not preoccupied with the road and traffic, it's an out-of-sight-out-of-mind situation as I watch my movie or edit my spreadsheet.
Just like it doesn't bother me when the train or bus is slow to get up to speed.
And electric vehicles are already relatively efficient in energy use even if driven badly, particularly if charged using renewables.
And, the next obvious step: private car ownership becomes a distant memory.
These are just some of the use cases. But also people who are fully capable of driving get advantages, particularly regarding commuting - instead of sitting in the car for an hour, one can rest, or if the job allows it, work on the way to the office.
Additionally: we, as humans, suck at driving. And whilst it's a valid question whether machines can currently do a better job at it, I'm glad that there's time and money spent towards a future without human drivers (apart from driving as recreational activity).
Still pretty interesting though.
When you leave the envelope of normal on a vehicle, it rarely increases the value.
Tl;DR the subject matter and the user-base of HN react in a way that brings out more negativity than each one could normally on its own.
I think your assessment is correct. Adding even more negativity to the mix, a lot of HN users are skeptical of Tesla in general and Autopilot in particular.
With autopilot in particular we know how that sausage is made.
[0] https://www.tesla.com/autopilot
But seriously I find the backlash so interesting and I think it is mainly for the reasons you describe. We all know how bug-laden much of the software that's shipped is, but it doesn't really matter because (for the most part) it's easy to iterate, issue an update - to the point that people who run un-updated software are considered foolish, etc etc. I think we should remember that actual _engineers_ are out there who design and make the bridges, cars and buildings that keep us alive who don't have this luxury. This guy seems like one of them and is clearly skilled (you can see that his workmanship looks good and he's taking care in these photos) and should be respected in my opinion.
In my opinion literacy at all is optional for 99.9% of automotive work but the kind of people who will pearl clutch over an enthusiast rebuilding calipers in their garage will probably not be all that pleased to learn that it's not uncommon for someone who can barely read the service literature to be servicing vehicles and I thought it was worth mentioning.
If we were talking about removing the safety valves on a steam locomotive so it could run hotter and faster you would see similar pushback from people who know what boiler explosions can do.
If he was testing at the track, at least he's only risking people that he knows.
If it gets something as small as a fender bender, and his insurance gets a whiff of this, you'll bet that they'll try hard to null-and-void the coverage.
Zero risk often achieves zero results - you have to balance your risk with other factors.
Also expecting it to be blacklisted by Tesla next week.
And I doubt there will be more than a handful of copycats, if that.
That said this guy is brave, not for the mod, but for posting pretty much all of the vin, the name of his car, and showing the color. Will not be hard for tesla to find it and blacklist charging, updates, parts, etc. Musk's ban hammer will be swift and harsh.
The idea of buying a car and being afraid to modify it because of how the manufacturer might react feels entirely dystopian to me.
They don't need to cover any damage done by the modifications, but they can't deny warranty coverage to anything else.
I'm not an expert, but I think the relevant law is the Magnuson-Moss Act.
Also, I wonder if all Tesla’s have a God Mode that unlocks at the sound of Elons voice or will come get him if needed.
So if he is ever running away from a villain in a major city he can access a getaway car anywhere he sees a Tesla.
A friend of mine is highly regarded engineer at one of electric car manufacturer. He supposed to be getting 8% corp discount for buying their vehicle. He doesnt so I asked him why since he’s so lucky. He told me eventually the day will come that local government or LE will force them to update their car’s OS to report continuesly on the speed you drive, similar to how tachometers work on trucks where cops stop you and can review your speed for the last 14 days you drove your rig nd either report you to your employee or guve you a retro-ticket for speeding. This update, he says will come and nothing you can do to stop it. I take my friend’s word very seriously although getting EV is as tempting as never before!
I think the car recording the historical data and the police using this as evidence is the much more likely future scenario.
Let’s not forget speeding fines are a revenue generator...
As for life saving, maybe, but I’d need to see a lot of evidence. Modern cars are vastly easier to drive and safer at speed than cars 40 years ago, but we still often keep largely the same speed limits. 155mph in your modern German luxury car practically feels serene.
https://www.businessinsider.com/ford-exec-gps-2014-1
If you try to fix a car and fail, you could end someone's life.
While I'm not against these kind of projects, to equate these things is ridiculous. Where is the line drawn? Vehicles are far more dangerous than your own PC.
Vehicles are dangerous to the few people around you, but a PC can be dangerous to millions.
How many people slap a turbo on their car without touching the wheels or brakes?
Nevermind if the upgrades or tune are sound to begin with.
1. https://www.youtube.com/channel/UCfV0_wbjG8KJADuZT2ct4SA
Tesla has _all_ the reasons in the world to find this car and take it out of service. On top of that, he killed the resale value.
Police as well. For sure it's going to be suck getting hit by this and having their insurance refuse to cover the damage