Tesla is preparing a significant update for its FSD (Supervised) system, focusing on two primary user complaints: navigating damaged road surfaces and inefficient route planning. Phil Duan, a lead software engineer on the Tesla AI team, confirmed that improvements addressing these issues are forthcoming.
The company aims to implement reliable pothole avoidance, a feature first promised in 2019 and listed in the release notes since FSD v14.3. While some current versions show sporadic avoidance behavior, the update intends to make this a stable, automated function that eliminates the need for manual intervention to protect wheels and improve ride comfort.
The navigation system will undergo a major overhaul to learn individual driving habits. By analyzing manual interventions and preferred routes, the FSD will adapt to the specific paths drivers choose for their daily commutes. This learning mechanism will also extend to parking, where the system will mimic a driver's specific habits, such as distance from curbs or garage walls.
Furthermore, Tesla is integrating the Grok AI model into the FSD system to enable natural language voice commands. This will allow drivers to issue instructions for turns, parking, or route changes without using the touchscreen. It remains unclear if these updates will arrive via minor patches or with the upcoming FSD v15, which is expected to feature a 10-billion parameter neural network.
