[dolphyvn]

Maybe a stupid question but is it possible with Mazda?

 

[sc489]

I certainly hope not judging by the software QA of the infotainment system!

 

[arathol]

Why (and how?) would the steering mechanism be connected to the radio??

 

[GerryB]

If there were a lane keeping assist function, then possibly.

Sent from my SAMSUNG-SM-G891A using Tapatalk

 

[lilstevie]

Why (and how?) would the steering mechanism be connected to the radio??

Because the CMU is connected to both the HS and MS CANBuses to get all the information that we can access. Just have a look at the speedometer app. All of that is information queried from the management computers.

One would hope that the drivetrain computers are firewalled from the CMU but without fully investigating its possible that they're not

 

[arathol]

Because the CMU is connected to both the HS and MS CANBuses to get all the information that we can access. Just have a look at the speedometer app. All of that is information queried from the management computers.

One would hope that the drivetrain computers are firewalled from the CMU but without fully investigating its possible that they're not

Exactly my point. Accessing the powertrain system sensor output via the OBD system is a far cry from actually controlling any of those systems. Add to that there is no reason at all why any control of those sensors or systems should be possible through the entertainment system. :surprise:

 

[[email protected]]

As there is a Lane Keep Assistance, which actually controls the steering wheel based on data known from the Camera Module, there probably is a way. Noone hase investigated it so far though.

 

[lilstevie]

Exactly my point. Accessing the powertrain system sensor output via the OBD system is a far cry from actually controlling any of those systems. Add to that there is no reason at all why any control of those sensors or systems should be possible through the entertainment system. :surprise:

It really isn’t, Charlie Miller has shown what can be achieved just by having access to the CANBus on a Jeep. https://youtu.be/ONDSAMfNGP0 If you want to see it in action

 

[arathol]

I have seen that, but there is more to it than what is shown. The car has to have some sort of similar functionality to accomplish such things, ie a feature that allows the ECU to operate the steering wheel. In the case of the Jeep, the way I understand it is this was accomplished by hacking the auto park software and removing the safeties that disabled the function when the vehicle was in forward gear. This one was also done by physically connecting the system to a laptop, requiring the hacker to actually be inside the car. An earlier more limited attempt was accomplished remotely using a wireless connection. The whole thing seems to be the result of faults in both the Sprint network and the Uconnect system used by Chrysler, which it seems had no real firewall protection. There is no reason to think that such a vulnerability exists in the Mazda system. Mazda at least seems to think so according to statements published back during the AIO debacle.

 

[lilstevie]

Which is fair, but many companies like to think that they've got no flaws in their software, when the reality is there are some massive ones and they can't be sure.

As I said, I would like to think that Mazda has firewalled it off so none of the critical systems share the same bus that features like LKA do, but unless someone experiments we wont know.

One benefit that we have vs the Jeep is that there is no cellular connection to our cars built into the CMU

 

[vic_bam85]

Which is fair, but many companies like to think that they've got no flaws in their software, when the reality is there are some massive ones and they can't be sure.

As I said, I would like to think that Mazda has firewalled it off so none of the critical systems share the same bus that features like LKA do, but unless someone experiments we wont know.

One benefit that we have vs the Jeep is that there is no cellular connection to our cars built into the CMU

In the cmu there is the command smdb-read (used by the speedometer app) which can be used to read information from the internal components (Limited). There is also the command smdb-write, but none of us have tried with it as it could damage the car or something. One of the available values is the steering wheel angle.

There is also the command dbus-send which can be used to read and also write values via dbus. This needs more research.

 

[lilstevie]

In the cmu there is the command smdb-read (used by the speedometer app) which can be used to read information from the internal components (Limited). There is also the command smdb-write, but none of us have tried with it as it could damage the car or something. One of the available values is the steering wheel angle.

There is also the command dbus-send which can be used to read and also write values via dbus. This needs more research.

Thanks, this is exactly what I’m trying to say. We don’t actually know if we can or not, and I’m sure a lot of us don’t exactly want to find out in case we do something that stops the car from working.

On another note, I do wonder if it would be possible to implement something like Forscan on the CMU to write asbuilt data. Certainly would be an interesting prospect if we could enable and disable features without needing to buy a HS/MS-CAN dongle

 

[gruuvin]

Steering input is not binary, but it's continuous with force and velocity, where angle changes over microseconds, and so voice control would not be good; like "turn left" is not sufficient: how far left?, how fast and when to stop turning left? When and how fast to turn back to the right to straighten back up? Ever try a driving simulator with keyboard buttons instead of a steering wheel? It's terrible, and your voice couldn't dictate the required input quickly enough. You would need a platform that can pretty much do all the driving for you with zero input (computer makes microsecond changes based on sensor input), and then you might be able to issue a voice command to "turn left at the next light."