Tesla has started rolling out its Supervised Full Self-Driving (FSD) v12.5 software globally, as announced by CEO Elon Musk.
Elon Musk has reiterated his confidence in Tesla’s ability to deliver a fully autonomous driving system without human oversight by year-end, cautioning that anything less would be unexpected for him.
Given the CEO’s track record of making identical predictions for the past five years, his current claims lack credibility.
Thereafter, our focus shifts to the timely delivery of Tesla’s Supervised Full Self-Driving (FSD) capability and its progress toward a fully autonomous system.
Musk may have overestimated the value of Tesla’s forthcoming software updates, 12.4 and 12.5. The autonomous vehicle, set to debut for 12.4, will notably eschew traditional steering wheels, instead relying on advanced navigation algorithms to enable efficient operation, potentially extending its range by a factor of 5 to 10 times between interventions.
Notwithstanding this, Tesla faced internal and external testing constraints, ultimately failing to roll out the update to the wider fleet.
As a substitute, Tesla will embark on a comprehensive global launch strategy spanning 12.5 months.
Musk stated during the discussion on FSD 12.5 and 12.6’s concluding month that it would require roughly 12 months’ worth of driving experience for the system to trigger its first intervention.
According to the CEO, Tesla plans to reintegrate freeway and road stacks in a future version of its software, V12.5, although this update does not appear to be included in the current 12.5.1 model.
As recently announced, the software update is currently available exclusively for HW4 vehicles, with Tesla working to optimize it for HW3 models pending further development. As Tesla draws closer to the limits of its Hardware 3 (HW3) capabilities.
Elon Musk revisited the issue with characteristic forthrightness.
Optimizing the code for execution on HW3 requires significant software development expertise. The validation process ensures that the data conforms to specific rules and criteria, thereby guaranteeing its accuracy and reliability.
While he estimates that HW3 cars might arrive approximately 10 days ahead of the initially predicted 12.5,
Electrek’s Take
I’m excited to scrutinize this claim, but my initial reaction is that a 12-month gap between interventions seems implausibly long, and I’d like to see more evidence before accepting such a significant lag time.
I’d be astonished if I didn’t receive an intervention within the first hour or a full year of driving.
Despite the progress made so far, I am increasingly uneasy about the prospect of multiple updates being required before an unsupervised system can be achieved, with potentially a dozen or more iterations still needed to reach that milestone. As Tesla has already devoted significant resources to refining the code for HW3, any subsequent updates will undoubtedly become increasingly challenging to implement.
I’m starting to harbour serious concerns that an unsupervised system will ever reach HW3 at its current pace?
