Peter,
Yes you're right. Looking at MINIX3 & other RTOS for vehicles & aerospace applications & railroads & other transport, I think the AUTOSAR focus is a mistake. More $$ & Euros are to be made with the needed support structures. The core should be OSS.
Elektrobit is focused on closed source AUTOSAR. I believe that leaves the market to not satisfying the safety concerns not being cross shared like is done on
kernel.org, The OSS culture promotes sharing a common code base through a Foundation company, testing lab, & working out needed government regulation.
I've been working on an Open Source
AUTOSAR,org . It defines lots of APIs. The auto & vehicle companies started this after several warranty crash & burn that cost on the order $14 B problems. One example BMW's early 2000s was part of that problem space.
There is one developer from Japan on GitHub who took the 1990s OTEK (sp?) as a start but hasn't gone very far as of yet. Given 100s of ECUs/Microcontrollers per vehicle OSS can help them communicate internally & externally.
RISC-V appears at 0.10 cents USD for lowend microcontrollers is a value. Ford here in Detroit is building it's own fabrication & battery manufacturing companies. U.S. companies are doing the same.
RISC-Vs are the low cost road ahead for the U.S., Europe, & Japan. Now what about OSS for core AUTOSAR's needed RTOS? MINIX3 recast into this heterogeneous distributed space can be a game changer up & down the stack.