Tesla updates FSD intervention menu to tag navigation errors

Teslarati reports that Tesla deployed Software Update 2026.2.9.9, adding a new Intervention Menu option for Full Self-Driving (FSD) disengagements. Per Teslarati, the menu now offers four categories: Preference, Comfort, Critical, and a new Navigation label that replaces the prior "Other" choice. Teslarati reports the change has been rolled out widely in recent days and is intended to let owners more accurately tag navigation-related interventions.

Teslarati describes the practical effect as cleaner, more granular labels for disengagement events. The article lists common navigation issues cited by owners, including incorrect speed limits, suboptimal routes, and routing that sends vehicles to a building's rear entrance instead of the main one. Teslarati frames these tags as inputs that can be fed back into neural networks and reinforcement learning systems to help isolate map- and routing-related failure modes.

Companies collecting human interventions typically rely on labeled event types to prioritise training data and debug model behavior. Better categorical labels reduce label noise and make it easier to aggregate examples for supervised training, metric calculation, and reward-shaping in reinforcement learning. For practitioners, clearer labels improve the signal-to-noise ratio when sampling failure cases for retraining or evaluation.

Editorial analysis: This is a low-effort UI change that can materially improve downstream data quality if adopted broadly. In autonomous-vehicle development, small shifts in telemetry or labeling practices often yield outsized benefits for triage workflows, dataset curation, and iteration velocity. The update does not, by itself, change model architecture or training regimen; it changes the upstream data pipeline by making one failure class explicit.

Watch for public reporting or changelogs that quantify the share of disengagements now labeled as Navigation versus the former Other bucket. Also track whether future patch notes or community posts link the label to specific map corrections, routing policy updates, or benchmarked improvements in on-road routing behavior. If third-party monitoring (forums, social media) shows rapid adoption of the label, practitioners can expect cleaner slices of navigation failures to surface in telemetry exports.