Not only do ICE car companies not issue OTA software updates, some use the same software for driver assist features for years in a row. I had heard good things about Subaru's EyeSight system and went to check out the 2019s Outback. I was floored to learn that the system is no smarter than it was in 2017, when this version came out.
Are they not learning from the millions of miles their customers have driven in the last two years? One would think that there would be at least some incremental optimizations (either for common cases or edge cases) to make the systems safer.
Some people actually prefer a car whose characteristics stay similar after they buy it. Sometimes these updates are beneficial, like when Tesla improved the braking performance by 20% after getting ragged on by Consumer Reports. But on the other hand I would be pissed if I bought a car that started trying to kill me after an OTA tweaks some weights in a neural net(e.g. https://www.reddit.com/r/teslamotors/comments/8a0jfh/autopil...)
I understand the risks of OTA updates, but it should be possible for them to release a 2019 model that has more nuanced algorithms than the 2017 model.
Another data-point is that I literally just went to talk by a researcher from Toyota's self-driving car group and they 1) mentioned some OTA updates they made on their auto-pilot features in the last several months. 2) talked about their data collection efforts from production cars- order of petabytes/day
I don't know much about subaru but other companies are definitely integrating advanced driver assist features/updates. No idea how they compare to Tesla's and in frequency of improvements but this is definitely a thing other companies are working on.
Glad to hear this. I recently test-drove the new RAV4 Hybrid, which just came out. I was unimpressed by their "lane tracing" system, which seemed noticeably worse than the CRV's equivalent.
I would be much more likely to pull the trigger on the Toyota (which gets much better gas mileage, due to the hybrid powertrain) if there was a hope of software updates.
I am hesitant to spend $30,000 on a vehicle whose driver assist features are currently subpar, and will only fall further behind over the lifetime of the vehicle. Makes me just want to lease, despite being generally opposed to leasing.
Even without LTE uploads, they could at least pull data down when vehicles are in for repairs, oil changes, etc. I realize they can't store infinite amounts of data on the vehicle, but they could definitely keep a little data about disengagements or other important events.
Are they not learning from the millions of miles their customers have driven in the last two years? One would think that there would be at least some incremental optimizations (either for common cases or edge cases) to make the systems safer.