Tesla’s ‘Supervised Full Self-Driving’ (FSD) program has been a topic of discussion for quite some time now. Despite CEO Elon Musk’s previous claims of significant improvements with the release of version 13 in December 2024, the data suggests otherwise. The promised 5 to 6 times increase in miles between disengagements compared to the previous version was not met, with only a 2 times improvement seen in the crowdsourced data.
Furthermore, Tesla’s HW3 vehicles are still stuck on version 12, with no plans to upgrade them to support unsupervised self-driving capabilities. The focus has now shifted to a “robotaxi” pilot program in Austin, Texas, which operates within a geofenced area and heavily relies on teleoperation support.
The recent release of version 13.2.9 has not shown significant improvements, with only 371 miles between critical disengagements. Musk has hinted at a more advanced model in the alpha stage with 4 times the parameters, but it will require further polishing before deployment, which is expected later this year.
However, history has shown that Musk’s timelines for FSD releases have often been delayed. The question remains – what will this long-awaited update bring to Tesla owners? The current focus on the Austin pilot program may lead to improvements in FSD for customer vehicles, but the promised unsupervised self-driving capabilities are still a distant goal.
It’s essential to manage expectations, as previous promises of parameter increases have not always delivered the expected results. While incremental improvements are welcomed, the ultimate goal of achieving level 4-5 unsupervised self-driving without teleoperation remains a challenge for Tesla.
In conclusion, while Tesla continues to work towards enhancing its FSD program, achieving true unsupervised self-driving capabilities for millions of vehicles remains a significant challenge. The focus on the Austin pilot program may lead to improvements, but scalability and reliability are key factors that need to be addressed before the promise of unsupervised self-driving can be fulfilled.
