Tesla adds new feature with FSD Beta v12 for more natural speed behavior

Tesla has recently introduced an innovative enhancement to its Full Self-Driving (FSD) Beta in the form of version 12.12, which promises a revolutionized driving experience. This cutting-edge update, which began reaching customers over the last weekend, is touted by the electric car manufacturer as its most reliable and precise iteration to date.

The cornerstone of FSD Beta v12 is its transformation of the driving stack for urban streets. The update pivots from relying on traditional coding, specifically removing over 300,000 lines of explicit C++ code, and instead, employs a unified, end-to-end neural network. This advanced network is the result of extensive training on millions of video clips, marking a significant leap in the development of autonomous driving technology.

Users of the new version have shared positive feedback, noting an enhanced sense of accuracy and confidence in their vehicle’s capabilities. Remarkably, some users report successfully completing complex drives without any need for human intervention, indicating the vehicle’s increased autonomy.

Yet, the upgrades extend beyond just refined driving dynamics. Among the notable additions is the “Automatic Set Speed Offset” feature. This functionality allows the vehicle to modulate its speed more effectively, adapting to various factors such as road type, traffic conditions, environmental influences, and the preset FSD Beta Profile (ranging from Chill to Assertive). The objective is to ensure the vehicle’s speed is in harmony with the surrounding traffic flow, even allowing for speeds above the posted limit if the situation permits, as per reports from Not a Tesla App.

Despite the advancements, Tesla’s CEO Elon Musk has acknowledged the need for continuous improvement. Specifically, he pointed out in December that the system requires further refinement, particularly in navigating areas with heavy precipitation, underlining the company’s commitment to enhancing the FSD technology further.