A long-running debate in the autonomous vehicle industry — whether cameras alone can safely replace human drivers, or whether additional sensors like lidar and radar are essential — has mostly been fought by engineers and executives. Now, lawmakers in New Jersey are attempting to settle the question through legislation.
A bill expected to face a vote later this year would require companies operating fully autonomous vehicles in New Jersey to use cameras combined with at least two other sensing technologies, most commonly lidar and radar. If passed, New Jersey would become the first U.S. state to codify such a hardware mandate into law, ahead of a nearly identical proposal still pending in neighboring New York.
The measure would effectively prevent Tesla's camera-only robotaxi system from operating in New Jersey unless the company alters its hardware approach. Tesla has invested heavily in the belief that artificial intelligence paired with cameras is sufficient for full autonomy, while most other major autonomous vehicle developers have taken a different path.
The Bill's Key Provisions and Safety Rationale
Democratic State Senator Andrew Zwicker, the bill's primary sponsor, emphasized that the legislation is not aimed at any single company. "This is not anti-Tesla," he told The Verge. "I'm pro-New Jersey safety."
Zwicker, a physicist who works at the Princeton Plasma Physics Laboratory, said he became convinced of autonomous vehicles' potential after riding in a Waymo robotaxi in Phoenix. He believes the technology could expand mobility, reduce traffic deaths, and improve transportation access. However, he argues it should be deployed cautiously in the nation's most densely populated state.
"At this point, I don't think the evidence is sufficient that a single sensor with software can handle situations that humans can," Zwicker said. "Can we get there? Maybe. But we're not there yet."
The proposal would establish a three-year pilot program governing the testing and deployment of fully autonomous vehicles in New Jersey. Companies would need to use multiple sensing technologies, report certain crashes, and obtain state authorization before launching fully driverless commercial services. They would also have to complete at least 50,000 miles of supervised testing in New Jersey without a major incident before removing human safety drivers.
The Industry Divide Over Sensor Technology
Elon Musk has long maintained that cameras paired with increasingly capable AI represent the best and most cost-effective path to autonomous driving. He has argued that humans navigate using vision alone, so sufficiently advanced AI should eventually do the same. Eliminating lidar and radar also significantly reduces hardware costs, making it easier to build robotaxis cheaply enough for large-scale deployment.
Musk has even contended that additional sensors can compromise safety by forcing software to reconcile conflicting data. "Lidar and radar reduce safety due to sensor contention. If lidars/radars disagree with cameras, which one wins?" he wrote on X last year. "We turned off the radars in Teslas to increase safety. Cameras ftw."
Most of the autonomous vehicle industry disagrees. Companies including Waymo and Zoox combine cameras with lidar and radar, arguing that each technology has distinct strengths and weaknesses. Cameras capture rich visual detail for recognizing colors, traffic signs, lane markings, and pedestrians, but can struggle in poor weather, darkness, or glare. Radar performs better in rain and fog and excels at measuring distance and relative speed of nearby objects. Lidar uses lasers to create detailed three-dimensional maps of a vehicle's surroundings, making it particularly effective at determining the shape and distance of objects.
Philip Koopman, a Carnegie Mellon electrical and computer engineering professor and autonomous vehicle safety expert, said camera-only systems may eventually become capable enough for fully autonomous driving, but he does not believe they are today. "Eyeballs are better than cameras for many reasons," he said, "and human brains are fundamentally more powerful than AI because we understand."
