Waymo Outperforms Tesla's FSD in San Francisco Test, Thanks to a Critical Error by Tesla

Tesla and Waymo, two leaders in autonomous driving technology, recently underwent a head-to-head test in San Francisco, offering a glimpse into the future of robotaxis. Business Insider's Lloyd Lee and Alistair Barr conducted the test using Waymo's Jaguar I-PACE equipped with the fifth-generation Waymo Driver and Barr's personal 2024 Tesla Model 3 running FSD Supervised software v13.2.8. The test was designed to evaluate the AI drivers' performance through a mix of city and suburban landscapes, including the challenging route from Twin Peaks to Chase Center and back. Waymo's robotaxis are fitted with a comprehensive suite of sensors, including five lidar sensors, six radars, and 29 cameras, designed to provide detailed environmental awareness. In contrast, Tesla's system relies solely on eight cameras, a much simpler and cost-effective setup. The Test Experience Waymo Ride The Waymo ride began at around 8:30 a.m. on a Thursday, with light to moderate traffic. The AI-driven vehicle demonstrated a balance of safety and assertiveness, such as confidently navigating yellow lights without taking unnecessary risks. One amusing and human-like moment occurred when the Waymo changed lanes to avoid a slow-moving car at a stop sign, showing a level of urgency that felt almost like a New York cab driver. Waymo’s AI driver handled the urban and suburban environments of San Francisco with ease, allowing the testers to feel comfortable and relaxed. Karp, a spokesperson for Waymo, explained that the Waymo Driver continuously plans its next moves, optimizing the route for safety and efficiency. Despite its high-tech sensor stack, the testers barely noticed they were in a robotaxi, which is a crucial measure of success. Tesla Ride Barr's Tesla, equipped with FSD Supervised, followed shortly after at around 10 a.m. under similar traffic conditions. Tesla's system delivered a generally smooth experience, handling lane changes and avoiding "phantom braking," a common issue reported in earlier versions of the software. The Tesla also excelled in highway driving, a capability that Waymo currently lacks, as the company is still testing and does not offer public highway rides. However, the Tesla made a significant error. Near the end of the ride, returning to Twin Peaks, the car turned onto a green bike lane and briefly drove in it before merging back onto the correct road. More critically, the Tesla ran a red light at the intersection of Twin Peaks Blvd and Portola Drive. The vehicle recognized the red light on the console screen but drove through it regardless, though it avoided hitting any cars or pedestrians due to the timing of nearby traffic lights. Verdict and Industry Insights Given the critical nature of the red-light violation, which would result in an immediate fail during a human driver’s license test, the testers concluded that Waymo emerged as the clear winner in this particular test. While Tesla’s FSD handled San Francisco’s hilly and winding roads almost as well as Waymo, the error in running a red light is a significant safety concern, especially considering that Tesla plans to roll out its unsupervised FSD robotaxis in June. Tesla CEO Elon Musk has often touted the company's reliance on real-world data from millions of vehicles on the road as the key to creating a more intelligent and versatile FSD system. However, the recent test highlights the importance of robust sensor technologies and the need for continuous testing and refinement. Waymo, with its advanced sensor stack and cautious approach, has taken a lead in the current robotaxi landscape. The company's focus on safety and reliability seems to have paid off in this non-scientific but telling test. As the race to fully autonomous ride-hailing services accelerates, it will be crucial for both companies to address these safety concerns and continue improving their AI systems to ensure a seamless and secure experience for users. Industry Evaluation Industry insiders highlight the critical importance of safety in autonomous driving. Running a red light, even in a controlled environment, underscores the potential risks involved in deploying unsupervised AI drivers. Waymo's multi-sensor approach and cautious programming seem to have given it an edge in this test, though Tesla's simpler, camera-based system shows promise and excels in certain scenarios. Tesla's upcoming unsupervised FSD rollout will be closely watched to see if it can match or surpass the performance and safety standards set by established players like Waymo. Both companies are pivotal in shaping the future of autonomous transportation, but the recent test suggests that Waymo is currently better poised to handle the complexities of urban driving.