Waymo Warns Cameras Alone Aren't Enough for Self-Driving Safety

As Tesla prepares to roll out its Cybercab robotaxis without steering wheels or pedals, Waymo is throwing down a serious challenge to the industry’s reliance on cameras alone. In a new blog post reflecting on over two million miles of autonomous driving, Waymo’s VP of Onboard Software, Srikanth Thirumalai, stresses that a purely camera-based approach falls short of ensuring passenger safety. While he doesn’t explicitly name Tesla or its CEO, the timing—just before Tesla’s robotaxi launch—makes the message unmistakable.

Waymo’s own fleet , which operates in multiple U.S. cities, relies on a suite of sensors including lidar and radar alongside cameras. Thirumalai argues that true autonomy demands redundancy and a deep understanding of environmental context. He points to lessons learned from heavy deployment, like the need for real-time adaptation and robust prediction models, which he says can’t be achieved by visual data alone. The post also hints at the importance of continuous learning from edge cases, something that requires more than passive image analysis.

This isn’t the first time Waymo and Tesla have diverged on technology philosophy. Tesla has championed computer vision as the sole solution, promising that its vehicles can achieve full autonomy with existing hardware. However, safety experts and regulators have repeatedly questioned the lack of redundancy in such systems. As the Cybercab launch approaches, Waymo’s warning raises fresh doubts about whether Tesla’s approach is truly ready for the roads.