Tesla starts supervised full self-driving v12.5 wide release

Tesla, the pioneering electric vehicle and clean energy company, has initiated the broad distribution of its latest Supervised Full Self-Driving (FSD) software, version 12.5, as announced by CEO Elon Musk. This development marks another step in Tesla's ongoing efforts to achieve fully autonomous driving capabilities.

Musk has once again expressed his optimism regarding the timeline for delivering an unsupervised self-driving system. He stated that he believes Tesla could fulfill this long-standing promise by the end of the current year, or at the latest, by next year.

However, it's important to note that Musk's track record with such predictions has been inconsistent. For the past five years, he has made similar claims about imminent breakthroughs in autonomous driving technology, which have not materialized as projected. This pattern has understandably eroded some of the credibility of his predictions in this domain.

Add WION as a Preferred Source

Given this history of unfulfilled promises, it becomes crucial to focus on Tesla's actual deliverables rather than future projections. Currently, the company is rolling out its Supervised Full Self-Driving system, and the progress of this technology serves as a more reliable indicator of how close Tesla is to achieving a truly unsupervised autonomous driving system.

Musk had previously generated considerable excitement about two upcoming software updates: versions 12.4 and 12.5. For version 12.4, he made bold claims, suggesting that it would operate without the need for steering wheel confirmation (often referred to as "steering wheel nag") and would be capable of driving "5 to 10 times more miles per intervention." However, Tesla encountered significant issues during both internal testing and limited external trials, leading to the decision not to release this update to the broader fleet of Tesla vehicles.

In light of these setbacks with version 12.4, Tesla has opted to proceed directly to a wide release of version 12.5. When discussing the capabilities of FSD versions 12.5 and 12.6 last month, Musk made an even more ambitious claim, stating that "it will take over a year of driving to get even one intervention." This assertion suggests a level of autonomy and reliability far beyond what has been demonstrated in previous versions.

Musk also indicated that with version 12.5, Tesla planned to re-integrate the highway and street driving capabilities into a single system, often referred to as "stacks" in software development terminology. However, it has been reported that this particular integration is not occurring in the specific 12.5.1 version currently being rolled out.

An important aspect of this rollout is that it is currently limited to vehicles equipped with Tesla's latest hardware version, known as HW4. The company is still in the process of optimizing the software for older HW3 systems. This limitation points to Tesla potentially reaching the computational limits of its HW3 hardware, which could have significant implications for the future development and deployment of more advanced autonomous driving features.

Addressing this issue, Musk commented that a considerable amount of software engineering effort is required to optimize the code sufficiently for it to run on HW3 systems. This optimised version needs to undergo separate validation processes to ensure its safety and reliability. Musk estimates that it will take approximately 10 more days before version 12.5 can be made available to vehicles running on HW3 hardware.

The author of the article expresses both excitement and skepticism about testing the new software. While eager to experience the improvements, they are highly doubtful of Musk's claim that the system could go a year between interventions. In fact, the author would be pleasantly surprised if they could use the system for even an hour without needing to intervene.

A significant concern raised by the author is the apparent gap between the current state of the technology and the promised unsupervised system. They estimate that at least a dozen more updates would be necessary to reach that goal. Given the difficulties Tesla is already facing in running the current code on HW3 hardware, the author anticipates that each subsequent update will pose increasing challenges for older hardware.

This situation has led the author to express serious doubts about whether an unsupervised system will ever be feasible on HW3 hardware, given the current pace of development and the emerging hardware limitations. This concern highlights the complex interplay between software advancements and hardware capabilities in the pursuit of fully autonomous driving technology.