Tesla Full Self-Driving Progress: A Leap Toward Autonomy?

Tesla’s Full Self-Driving (FSD) software is the linchpin of its robotaxi ambitions, and its progress has been both impressive and contentious. As of April 2025, FSD version 13 refines neural network-driven navigation, achieving smoother turns and fewer disengagements than its predecessors. Version 14, slated for mid-2025, promises to integrate audio cues and advanced transformers, potentially enabling “unsupervised” autonomy by June—crucial for the Austin robotaxi launch. Tesla claims internal tests in Fremont show Model Ys autonomously handling complex routes, a leap from the Level 2 capabilities of 2020’s beta.

Yet, challenges persist. Critics highlight a 71-mile average between critical disengagements, far behind Waymo’s 17,311 miles, and the NHTSA’s probe into FSD-linked crashes underscores safety gaps. Tesla’s camera-only approach, eschewing LiDAR, has sparked debate: it cuts costs but struggles with edge cases like fog or glare. On X, engineers praise FSD’s data advantage—millions of real-world miles from Tesla owners—while skeptics argue it’s still a “science experiment.” If version 14 delivers, Tesla could silence doubters; if not, the June timeline risks another delay, testing investor patience and Musk’s credibility.