Waymo Challenges Tesla, Emphasizing the Need for More Than Just Cameras

As Tesla prepares to roll out its steering wheel and pedal-less Cybercabs, Waymo is sounding the alarm about Elon Musk’s approach to autonomous driving. Srikanth Thirumalai, Waymo’s VP of Onboard Software, doesn’t directly name Musk or Tesla in a recent blog post titled “10 AI Lessons from Driving 200+ Million Fully Autonomous Miles,” but his message is unmistakable: Tesla’s system falls short for true autonomous driving.

“Cameras are incredible, but they aren’t enough,” Thirumalai states. He emphasizes that after accumulating over 200 million real-world miles, the evidence is clear: achieving safe, fully autonomous operations at scale requires more than just cameras. Waymo combines cameras with lidar and radar to create a comprehensive, redundant view of the world that no single sensor can replicate.

Musk, however, has a different perspective. Over the years, he has openly criticized lidar, labeling it “a crutch” and suggesting that companies relying on this costly technology are “doomed.” His argument hinges on the idea that humans primarily rely on their vision for driving decisions, so why shouldn’t autonomous vehicles do the same?

Recently, Tesla has doubled down on this stance. During an earnings call, Tesla’s head of AI, Ashok Elluswamy, took aim at competitors using a multi-sensor approach. “Historically, the so-called experts have always claimed that you need lidars, radars, HD maps, and the entire kitchen sink to drive safely,” he asserted. “Here we show that such is not true. You can have safe, comfortable, and affordable autonomy with just cameras. This record should be a huge validation of Tesla’s entire AI approach.”

Tesla’s camera-only strategy, however, may limit its access to certain markets, especially as policy discussions evolve. For instance, New Jersey is considering legislation that would permit robotaxis only if they incorporate multiple sensors, which would exclude Tesla.

Thirumalai also highlights other distinctions between Waymo and Tesla. While Musk dismisses high-detail maps as “a really bad idea,” Waymo views them as a valuable “prior.” Thirumalai explains that Waymo uses HD maps to enhance its validation process, enabling a fully autonomous service from the very first trip. “As we drive, we treat our maps as another input — like our sensors, but acting as a mental memory,” he notes.

Musk argues that autonomous vehicles should operate without maps, as roads can change, lanes can be repainted, and construction zones can emerge. He believes that a system capable of instant analysis should rely less on maps, asserting that HD maps could hinder the development of such a system.

Waymo, on the other hand, is not resting on its mapping achievements. The company employs an AI-driven mapping system that ensures its maps are “continuously updated,” providing vehicles with a reliable, high-fidelity reference for complex maneuvers. In essence, while mapping is beneficial, it’s not the sole solution.

Thirumalai elaborates on additional principles at Waymo, such as the value of closed-loop simulation for identifying edge cases and the effectiveness of Vision-Language Models as reasoning tools. He concludes with a pointed critique of Tesla: a fully autonomous system cannot be built on a Level 2 driver assist framework.

“The final lesson is that there is no substitute for actual fully autonomous miles. Simply improving a driver-assist system (L2) for full autonomy is a false summit. True L4 maturity can only be safely achieved by a purpose-built system, validated on closed courses and hardened by the uncompromising experience of driving without a human in the car.”

This statement directly challenges Tesla’s approach, which posits that with enough data, miles, and training, its advanced driver assist systems, Autopilot and Full Self-Driving (Supervised), will eventually evolve into Level 4 capable, fully autonomous platforms.

Currently, Tesla’s FSD system is classified as Level 2, meaning drivers must remain vigilant and ready to take control at any moment. In the event of a crash while FSD is active, Tesla typically places the blame on the driver, citing terms of service that require constant monitoring. However, this contrasts with how Musk and other executives present FSD. The company even had to add “Supervised” to its marketing after regulators raised concerns about misleading claims.

So, who’s in the right? Waymo or Tesla? Waymo operates fully autonomous vehicles in 11 cities, completing 500,000 paid trips weekly. Tesla has its own robotaxis, some supervised by human safety drivers, but an increasing number are unsupervised. The company is on the verge of launching its dedicated Cybercab, though its initial rollout remains uncertain.

Despite years of boasting about its autonomous capabilities, Musk is clearly trailing behind Waymo. While it’s unlikely he will pivot to embrace lidar, he is undoubtedly aware of the gap between his company and Alphabet’s robotaxis. How he plans to address this challenge will become clearer in the coming months.