Tesla’s Full Self-Driving system, version 14, faced a demanding real-world challenge across some of Los Angeles’s most unpredictable roads. The test included steep mountain routes with blind corners, narrow one-lane stretches, and chaotic intersections without lane markings. Traffic conditions were pushed to the limit, featuring large garbage trucks, unexpected pedestrian movements, and the notoriously difficult on-ramp to the 110 Freeway—where acceleration lanes are nearly nonexistent. These scenarios represent extreme edge cases that stress even advanced driver-assistance systems.
The results revealed a mix of impressive adaptability and surprising limitations. At times, the vehicle made decisions that mirrored human intuition, such as adjusting speed for tight turns or yielding appropriately. However, it also exhibited confusion in complex situations, including misjudging gaps during merging and hesitation at ambiguous traffic patterns. These inconsistencies highlight the ongoing challenges of deploying autonomous systems in unpredictable urban environments, especially where infrastructure and behavior don’t conform to ideal conditions.