Ex-Tesla Autopilot Head Calls Vision-Only a 'Weird Religious Stance'

Sterling Anderson, the former head of Tesla’s Autopilot program and current Chief Product Officer at General Motors, has criticized Tesla’s reliance on camera-only autonomous driving, describing the decision to drop radar as "a weird religious stance that ignores physics."

What Happened

In an interview with The Verge published on October 27, Anderson stated that scaling back to vision-only systems creates significant challenges. "If you scale back to vision-only, physics suggests you’ll be in trouble in certain situations," Anderson said. He argued that radar is inexpensive and that cameras cannot see through dense fog or measure instantaneous velocity as effectively as radar. Anderson also noted that lidar helps camera systems improve faster when the two are trained together, countering Tesla’s position that cameras and fleet data are sufficient.

Anderson’s comments come from a unique position; he directed Tesla’s Autopilot during its formative years in 2015 and 2016 before co-founding Aurora. He joined GM in 2025, where he is overseeing a roadmap that includes the Cadillac Escalade IQ, which is due to debut GM’s "eyes-off" system in 2028 using lidar, radar, and cameras. Anderson also criticized the use of "eyes off" marketing language, arguing that highly capable driver-assist systems already push drivers toward complacency.

Why It Matters

The critique highlights a growing divergence in autonomous vehicle strategies as Tesla pushes camera-only Cybercabs into driverless service in Austin. Tesla began removing radar from Model 3 and Model Y vehicles in May 2021, followed by ultrasonic sensors in 2022. The transition to Tesla Vision introduced limitations such as an Autosteer speed cap of 75 mph and disabled features like Smart Summon at delivery.

Regulatory scrutiny has intensified around these changes. In March, the National Highway Traffic Safety Administration (NHTSA) upgraded its investigation into Tesla’s Full Self-Driving (Supervised) system in reduced-visibility conditions to an Engineering Analysis covering 3,203,754 vehicles. The probe cites nine crashes, including a pedestrian fatality in November 2023. NHTSA found that FSD’s camera degradation detection failed to identify common roadway conditions impairing visibility, such as sun glare, fog, and airborne dust, often alerting drivers only immediately before crashes. Anderson’s emphasis on sensor redundancy underscores the safety risks associated with relying solely on optical inputs in adverse weather.

The Bottom Line

Tesla’s vision-only approach faces continued criticism from former leadership, with Anderson arguing that cheap, reliable sensors like radar and lidar provide necessary redundancy. While Tesla defends its first-principles engineering stance, data from NHTSA investigations and the operational scale of competitors like Waymo—who use lidar, radar, and cameras—suggest that multi-sensor systems may offer greater robustness in complex driving environments.