Comment Re:Betting The Company Gone Wrong (Score 1) 28
This is true, and I didn't mean to imply the entire platform was bad and always would be. The theory is great, and it should have become a model for the industry if it were well executed. There were fundamental issues with transitioning from hard realtime automotive microcontrollers in a domain based architecture to the soft realtime world of QNX, task scheduling, and the realities and limitations of Nvidia AGX Xavier in a centralized architecture that hurt a lot. Real QA was largely absent, as it was deemed more important to have "green lights" on automated test system dashboards. Because the cars were so late in development with major hardware changes happening too late in the program, combined with software that was not safe for the road, it was extremely difficult for engineers to actually drive the cars. When early production began and engineers were allowed to lease the cars, tons of bug reports came in with many serious issues, but the cars were still deemed ok to ship to customers, with executive and engineering management fully aware of the situation.
Hopefully the next platform, SPA3, a zonal architecture, will be better, even if they don't have Luminar lidars.