Tesla's self-driving software fails to detect trains at crossings

In North Texas, several Tesla owners have reported alarming incidents involving the company's Full Self-Driving (FSD) software, particularly concerning its inability to recognize and safely navigate railroad crossings. These incidents have raised significant safety concerns among drivers and experts alike. One notable incident involved Italo Frigoli, who narrowly avoided a collision when his Tesla, operating in FSD mode, failed to detect an oncoming train at a crossing. Video evidence corroborated his experience, and tests conducted afterward indicated that the software continued to malfunction at that location, with similar reports coming from at least six other Tesla drivers. The National Highway Traffic Safety Administration (NHTSA) has acknowledged awareness of these incidents, emphasizing their ongoing communication with Tesla to ensure the safety of road users. The agency has pointed out the potential dangers arising from these software failures, with experts warning that it could lead to fatal accidents. Phil Koopman, an associate professor emeritus of engineering, indicated the gravity of the situation by saying it's dangerous if the software cannot stop at rail crossings. Such technology is viewed critically by experts, suggesting that the existing training data used by Tesla engineers might lack sufficient examples of train crossings, contributing to the failures observed. The rail industry has also voiced its concerns regarding autonomous vehicle safety at railroad crossings for years, exemplified by warnings from the Association of American Railroads in 2018. They stated the importance of recognizing different signals, such as locomotive headlights, horns, and bells, especially since not every crossing is equipped with gates or flashing lights. In light of these findings, Musk's promotion of FSD as the future of transport continues to come under scrutiny as critics argue that the system, classified only as a Level 2 automation requiring constant human oversight, is overhyped and potentially misleading. With the number of reported incidents increasing and video evidence emerging online, detailed investigations into the software's performance at rail crossings may become a priority. As Tesla mechanisms evolve in a rapidly changing automotive landscape, Tesla owners and safety experts alike stress the need for improvements in FSD software reliability, especially in critical situations like railroad crossings where human lives are at stake.