Hacker News new | past | comments | ask | show | jobs | submit login
GM Backed "Open Source Protocol" for Connecting Vehicle Apps and Services (eclipse.org)
30 points by super_linear 6 months ago | hide | past | favorite | 41 comments



I don't know that I trust this. I was in a GM vehicle and on start the entertsinment unit showed a URL where you can access open source licensed code used in the car. Went to the URL and it was a dead link. I probably should've notified the EFF or something but I didn't. These companies don't understand or respect FOSS.


> The purpose of this project shall be to provide a transport agnostic, layered communication protocol that builds on top of existing automotive and Internet standards, from the mechatronic layer (between ECUs, VMs, etc…) up to the Cloud, enabling a connected software defined vehicles.

"Up to the cloud"

There is no need for a car to "up to the cloud". Just give me a local Bluetooth network that reports back to me engine ECU reports, statistical logistical reports and driver stats. Because as soon as you leave the ground to "reach the cloud" that's where open-source stops.


enabling a connected software defined vehicles.

The end-goal is, not surprisingly, to take away ownership and control. "Open source" is just a distraction, a way to paint this in a trendy positive light. Lots of things that are truly open source are actually user-hostile too. Whether the source is open doesn't really matter here.


Meanwhile my SO really enjoys starting the car's climate control when she's about to leave her friend, the restaurant or similar, way out of Bluetooth range.

Especially during the winter when it's -15C or below it's really nice to step into a warm car compared to an ice cold car.


It doesn't have to be Bluetooth. We had local wireless remote start and control solutions long before "cloud" or even BT or WiFi was a thing.


Add LoRA support to phones and you could have it miles away.

But that would not require a subscription.


Heck, put LoRA and Bluetooth Low Energy in the remote, then your phone can send this message through BLE and the remote will relay it to the car seamlessly from significant distance.


I don’t want to sell all my data and privacy to car companies so I can have a warm car from a distance.

If you want to make that trade you guys go ahead. Me: bad trade.


And I find it a bad trade NOW. I can't imagine what it's gonna be like in 10 years once they don't push security updates and the Kia Boys can use years old security bugs to steal your car.


Oh yea. They toute “cars as technology platforms” but these are the organizations who couldn’t even deliver updates over-the-air until Tesla shamed them. The attack surface of entire product lines that have huge utility but no security is going to be epic.

Just look at this garbage. Not one mention of any of the issues or security problems these companies are condemning themselves to, not to mention the consumers…

https://www2.deloitte.com/content/dam/Deloitte/de/Documents/...

Vehicles can last 10-20 years, but updates from manufacturers top out at 5-7 and are almost exclusively closed source and lawsuit / litigation heavy and willing.

It is kill box territory.


proc HeatUpCar {$temperature} { switch $temperature { -15c { activateHeat } } }

Not a hard feature to program


I think the idea is that you can start the heater a few minutes before you head to your car rather than setting a return time before you leave, or leaving the car's heater on the entire time.


Hmm, touché.


Fortunately there is 30 year old tech that does exactly this without any modern app-tier enshittification.

You can start a car from 2km away by pushing a button on the key ring. This is a solved problem. That doesn't involve a smartphone.


How much range when the car is in a parking garage and you are in a condo, restaurant or similar, ie both surrounded by concrete?

Besides an app allows much more than a simple on off.

Not saying you should have to sell your privacy for ut, but there are some potential positive things by having the car connected to the cloud. So if there's a way to do that without the shenanigans then that might be good.


Unfortunately that battle appears to be over. Automakers are never going to surrender their toehold in our cars without a fight.

https://news.ycombinator.com/item?id=38234716


At this point the horse is out of the barn. People want a connected center screen. Figure out how to do it in an open way if you must, but telling people not to want it doesn't work.


People want tactile hard buttons, not a "connected center screen".

Who started this myth?


Marketing, of course. People are being told what they should want, which conveniently lines up with cost savings from the companies themselves..


No its actually that those cars sell better. Specially if you look to China its even more true. Even in the US pickup trucks turn into luxery cars with lots of tech and gadgets.

I love how tech people always believe that what they dont like nobody likes and consumers are forced.

Most people simply don't give shit if things go over local networks or clouds. People like opening an app and seeing if somebody is stealing from their car. To prewarm the car from whereever they are. And so on and so on.

And its not actually what traditional car companies want. In fact, they are in panic mode because new companies like Tesla and car makers from China have interduced these things and costumers prefer prefer it.

So now they are rushing to catch up and often doing it badly.


They want it to be a phone so bad.


There's some previous submissions ~6 mo ago but not much traction.

Since it's not named in the title it's: Eclipse UProtocol. Protobuf & CloudEvents. Url-based, with different devices ("UDevice") each getting their own dns name ("UDomain"). Somewhat transport agnostic. Http2, amqp, mqtt, dds (used some in ROS robotics os).

This makes me rather miss Webinos, which was a connected is with pretty interesting end user privacy systems kind of built in that let the device kr car talk to a cloud, but a user defined cloud.


After the failure of pretty much all embedded automobile platforms like microsoft sync that people really don't want yet another crappy middleware platform, the automotive industry is trying this shit again removing android auto and apple carplay after people actually got what they want.

Why? Auto vendors realized they lost control through their ineptitude of the user interface, and want it back. Problem is no one wants what they have to offer, that is a sub-par experience in every way to simply letting them use their phone as a screen in their car already there.

Automobile vendors realized they screwed up and lost real estate while they were still screwing with Microsoft, Clarion, all the other WinCE vendors and few others longer than they should, and are simply going to piss everyone off trying to take it back again.


That hasn't been my experience with a 2021 RAV4. The Toyota system is much more reliable then Carplay. Which I've heard is itself more reliable than android auto.

I imagine partly because it uses the physical buttons located around the screen, partly because automakers have much more stringent requirements for software validation.

e.g. Maybe 1 out of 1000 times the built in maps app crashes, gives an error, or doesn't respond to an input. That ratio is more like 1 in 10 for Maps in Carplay.


If Toyota integrated CarPlay correctly, it wouldn’t be crashing. Instead, they prioritize their inferior system, which only leads to better experience because the other one is gimped.


That's possible, but I haven't seen any evidence that Toyota's integration is notably worse then other automakers.


They didn't even start supporting it until the past couple years, so of course it's going to be worse. They're years behind on institutional know-how and infra


> They didn't even start supporting it until the past couple years, so of course it's going to be worse.

What's the reasoning behind this? Apple itself tends to be late in supporting a lot of things, but when they do it's more polished then competitors.


Toyota has massive scale, is very successful, and is japanese, and thus has extreme inertia on change, even in software where such things 'should' be faster. Their existing infotainment has been considered fine and functional for years, and Toyota likely didn't start seriously considering it till consumers were already demanding it.


Did you mistake my comment with a different comment chain?

I was addressing the parent's claim that the Carplay integration is 'much worse' then other automakers.


I've never experienced a "crash" in Maps or Google Maps, whether on carplay or on the phone. Maybe there is something wrong with your car/phone?

Carplay & android auto do have issues connecting at first but the apps themselves never seem to crash.


I said 'crashes, gives an error, or doesn't respond to an input'?

Have you never had an unresponsive screen before?


I’ve used carplay with a iphone 12 nearly every day for 2 years and never had issues with it crashing or becoming unresponsive. Stability issues definitely makes me think it’s a problem with your setup.


It sounds like you are an outlier then, there are many many videos on youtube showing Carplay to not respond 100% of the time.


In Toyotas?


Yes? Even if only 1% of Toyota owners total post videos of their screen and only 1% of those videos capture carplay interactions, that's thousands of videos total.


Google Maps crashes all the time on my Pixel phone, and I hear similar complaints from friends of mine on Samsung and Motorola phones. Note that this is without using Android Auto.

Count yourself lucky if you have a stable, crash free Google Maps experience!


Feel like google maps runs better on iOS.


Just wait until you want to drive to a newly built neighborhood and it isn't loaded into your maps so you can't do normal GPS navigation there.


I am more than happy with my infotaiment system, don't need Android or iOS messing up with it, bluetooth audio is more than enough.


There's some previous submissions ~6 mo ago but not much traction. https://hn.algolia.com/?q=%22uprotocol%22

Since it's not named in the title it's: Eclipse UProtocol. Protobuf & CloudEvents. Url-based, with different devices ("UDevice") each getting their own dns name ("UDomain"). Somewhat transport agnostic. Http2, amqp, mqtt, dds (used some in ROS robotics os).




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: