Waymo Says Cameras Aren't Enough: The Self-Driving Sensor War Heats Up

View original article on TekMag

Waymo's VP of Onboard Software published a blog post on August 20, 2026 that drew a clear line between Tesla's camera-only autonomous driving approach and the multi-sensor strategy powering Waymo's 200 million autonomous miles.

Srikanth Thirumalai, Waymo's vice president of onboard software, published "10 AI Lessons from Driving 200+ Million Fully Autonomous Miles" on the Waymo Waypoint blog. The post doesn't name Tesla or Elon Musk directly, but the target is clear: a direct challenge to the camera-only philosophy behind Tesla's Full Self-Driving (Supervised) program and the upcoming Cybercab robotaxi.

Tesla is preparing to launch its steering-wheel-free Cybercab on September 3, 2026, in Austin. Waymo already operates fully autonomous vehicles in 11 U.S. cities, completing 500,000 paid trips per week. The blog post vindicates Waymo's sensor fusion strategy and argues that Tesla's approach relies on a false premise.

Key Takeaways

Waymo's VP says cameras alone cannot achieve safe, fully autonomous operations at scale, a clear reference to Tesla
Waymo uses cameras, lidar, and radar to create redundant sensor coverage for edge cases like heavy rain and fog
Tesla's FSD remains classified as Level 2, requiring constant driver monitoring
New Jersey is considering a robotaxi bill that would ban Tesla's camera-only vehicle from the state
Waymo operates 500,000 autonomous trips weekly across 11 cities with 200+ million miles logged

Cameras Alone Cannot Handle the Real World

Thirumalai's opening line reframed the autonomous driving debate: "Cameras are incredible, but they aren't enough."

For years, the industry has split into two camps. Tesla and Musk argue that human drivers rely primarily on vision, so autonomous vehicles should too. Waymo, Zoox, and Motional take the other side, equipping their vehicles with lidar, radar, and cameras — a multi-sensor suite that provides redundancy when one modality fails. However, the Waymo blog presents this as the only path to reliable autonomy.

Waymo's argument rests on data. The company has driven over 200 million miles fully autonomously, a distance unmatched in the industry. That mileage puts it in a small club of operators with real-world validation at scale.

The blog post explains how each sensor type covers the blind spots the others miss. Lidar captures 3D geometry with millimeter precision. Cameras read street signs and detect traffic light colors. Radar tracks velocity and sees through heavy rain, fog, or dust that obscures cameras. The combination creates a "rich, redundant world view that no single sensor can replicate."

HD Maps: The Other Fault Line

Beyond sensors, the Waymo blog targets Tesla's mapless philosophy. Musk has called HD maps "a really bad idea," arguing that roads change constantly. Lane markings shift, construction zones appear, potholes form. A vehicle should navigate based on real-time perception, not pre-rendered geography.

Waymo takes the opposite view. HD maps serve as a "powerful prior," acting like mental memory for the vehicle. They jump-start the validation process so the system can operate safely from the first trip, not after millions of miles of learning.

Thirumalai emphasized that Waymo's maps aren't static. An AI-driven mapping system ensures they're continuously updated. The vehicle treats the map as another input, like a sensor, but providing context rather than raw perception.

HD maps require constant maintenance and cannot anticipate every road change. Tesla's mapless approach promises broader geographic coverage but demands more from the perception stack in unfamiliar areas.

The L2 to L4 "False Summit"

The blog's most direct critique appears in lesson ten: "There is no substitute for fully autonomous experience."

Thirumalai wrote that simply improving a Level 2 driver-assist system for full autonomy is a "false summit." True L4 maturity requires a purpose-built system validated on closed courses and hardened by driving without a human in the car.

This is an unambiguous reference to Tesla's trajectory. Tesla's Full Self-Driving (Supervised) is classified as Level 2. Drivers must monitor the road at all times and be ready to take control. Tesla added the word "Supervised" to its marketing after regulator complaints that the system misled customers about its capabilities.

Musk has repeatedly claimed Tesla's FSD will evolve into full autonomy given enough data and miles. Waymo argues architecture matters more than mileage count. A system designed for supervised driving has different safety assumptions than one built from the ground up for unsupervised operation.

Tesla's Defense

Tesla isn't backing down. On a recent earnings call, Ashok Elluswamy, Tesla's head of AI, pushed back hard against the 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," Elluswamy said. "Here we show that such is not true. You can have safe, comfortable, and affordable autonomy with just cameras."

Musk has made similar comments for years, calling lidar "a crutch" and predicting that companies relying on the technology are "doomed." His argument rests on the observation that humans drive using eyes and brains, so machines should be able to do the same. This camera-only philosophy has influenced AI hardware trends across the industry, as explored in our coverage of AI-driven development tools.

Waymo counters that humans also benefit from peripheral vision, depth perception, and instinctive risk assessment that cameras alone cannot replicate at machine speed. More importantly, humans make errors. An autonomous system needs to be more reliable than a human, not just comparable. Tesla's camera-only approach, which relies heavily on AI compute from NVIDIA, has been praised by CEO Elon Musk as the path to affordable autonomy, while critics point to its limitations in adverse weather.

Tesla remains unusual in its commitment to cameras only. The vast majority of autonomous vehicle operators — Waymo, Zoox, Motional — use multi-sensor suites. As The Verge reports, Tesla is "mostly alone" in betting everything on vision.

Regulatory Pressure Is Building

The sensor debate is moving into law. New Jersey is considering a robotaxi bill that would require autonomous vehicles to include cameras plus at least two additional sensor types. That requirement would effectively ban Tesla's camera-only Robotaxi from operating in the state. The bill also mandates 50,000 miles of supervised testing, crash reporting, and state authorization before commercial launch, with a three-year pilot period.

Tesla responded aggressively. The company flooded New Jersey state legislator Zwicker's office with approximately 4,000 protest emails in a single day, according to The Verge. The lobbying effort suggests Tesla views multi-sensor mandates as an existential threat to its robotaxi ambitions.

Other states may follow New Jersey's lead. If regulations require lidar or radar, Tesla's cost advantage from a simpler sensor stack disappears, or its vehicles simply cannot operate in those jurisdictions.

The Numbers Tell the Story

Scale separates Waymo from its competitors. The company's 200+ million autonomous miles cover 11 U.S. cities. The system completes 500,000 paid trips weekly. That reflects real-world operation at a scale no other company has matched.

Tesla's robotaxi effort remains smaller. Supervised driverless Model Y vehicles have run in Austin, Dallas, Houston, Miami, Orlando, and Tampa since June 2025. An increasing number operate without safety drivers, but the fleet is limited compared to Waymo's footprint. The Cybercab — a two-seat, pedal-less vehicle built from the ground up for autonomy — launches September 3, 2026, in Austin, according to Motor1.

The question is not just which approach is safer. It is which can scale faster across different geographies. Tesla's camera-only system is cheaper to produce, but Waymo's multi-sensor setup has proven more reliable in complex urban environments.

Looking Ahead

The autonomous driving war is entering a decisive phase. Tesla's Cyberab launch in September will test whether a camera-only vehicle can operate safely without human oversight at scale. Waymo's response will likely come through expanded service areas and continued mile accumulation.

Regulators are watching closely. New Jersey's proposed bill signals that policymakers may increasingly tie safety requirements to sensor diversity, not just operational metrics. Other states may adopt similar frameworks.

For now, Waymo's blog post is a confident statement from the market leader. Tesla's Cyberab launch will either close the gap or confirm Waymo's argument, whether the company intends it to.

Conclusion

Waymo's August 2026 blog post draws a clear line: the multi-sensor approach has won the evidence, and Tesla's camera-only bet remains unproven at scale. With the Cyberab launch just weeks away, the coming months will determine whether Tesla can close the gap or whether regulators follow New Jersey's lead and codify multi-sensor requirements into law.

What are your thoughts on the sensor debate? Share your opinion in the comments below.

Frequently asked questions

Q: What is Waymo's stance on Tesla's camera-only autonomous driving approach?

Waymo's VP of Onboard Software Srikanth Thirumalai stated in an August 20, 2026 blog post that "cameras are incredible, but they aren't enough." Waymo argues that safe, fully autonomous operations at scale require a multi-sensor approach combining cameras, lidar, and radar for redundancy and comprehensive perception.

Q: How many autonomous miles has Waymo driven?

Waymo has driven over 200 million fully autonomous miles across 11 U.S. cities, completing approximately 500,000 paid trips per week. This scale of unsupervised operation is unmatched in the industry.

Q: What is the difference between Level 2 and Level 4 autonomy?

Level 2 autonomy requires constant human supervision — the driver must monitor the road and be ready to take control at any time. Level 4 autonomy means the vehicle can handle all driving tasks in defined conditions without human intervention. Tesla's FSD is classified as Level 2, while Waymo's system operates at Level 4 in its deployed service areas.

Q: When is Tesla's Cybercab launching and where?

Tesla's official Cyberab launch event is scheduled for September 3, 2026, in Austin, Texas. The two-seat vehicle has no steering wheel or pedals and relies on Tesla's AI4 computer for autonomous operation.

Q: Could New Jersey's robotaxi bill ban Tesla vehicles?

Yes. New Jersey is considering a bill that would require robotaxis to include cameras plus at least two additional sensor types, effectively mandating lidar and radar. Since Tesla's planned Robotaxi uses cameras only, the legislation would prevent the vehicle from operating in the state. Tesla lobbied against the bill by sending approximately 4,000 protest emails to a state legislator's office in one day.