What self-driving cars need

Self-driving Tesla electric vehicles could be on the road sometime next year under CEO Elon Musk’s vision to bring financial success to his upstart company.

But critics have called that vision faulty, saying Musk can’t deploy autonomous vehicles safely without Lidar sensors that send out laser light beams to detect objects. Tesla’s system relies on eight cameras, short-range sonar sensors and front-facing radar.

Musk, at an April event to detail his system, called Lidar expensive and unnecessary. Tesla, he said, had developed a high-speed computer and software that would use data and artificial intelligence to safely navigate the roads. During the presentation, he called Lidar a “fool’s errand” and “frigging stupid,” mocking most other companies developing autonomous vehicles with the laser system.

Marta Hall, president of San Jose, California-based Lidar marker Velodyne Lidar Inc., answers questions about Tesla and Lidar.

How would you respond to Elon Musk’s assertion that Lidar isn’t needed?

I don’t think he knows what he’s talking about. He doesn’t understand how Lidar works to make driving more safe. I don’t think when he talks about autonomy it’s truly autonomous. I think what he’s talking about are advanced features that are kind of cool but they’re not fully autonomous and they’re not fully safe as he describes them. He’s had leeway because Elon has a loud voice and so people listen to him. But in this case, when he’s talking about Lidar, it doesn’t seem that he knows the science that’s in the instrument, and he thinks he can drive a car autonomously with just cameras and radar, which is dangerous.

What does Lidar offer that a camera and radar don’t?

Lidar is 3-D vision, and cameras and radar are two-dimensional. Lidar can see at night when camera vision is very poor. The majority of accidents happen at night.

What’s so dangerous about Musk saying he has full self-driving without Lidar?

Customers will buy his car, and they will take a nap in the car. Cameras fall short because of optical illusions, just like the human eye can be fooled. Lidar can’t be fooled in the same way. Lidar sees in measurements. It sends out light beams that actually measure distance. Cameras, on the other hand, have to make assumptions about how far away objects and other cars are. So if you want to avoid accidents, which would you rather rely on, assumptions or math and measurements?

Musk could say you’re just trying to sell your product.

It’s more than that because none of the other car companies wants there to be more accidents. He’s fooling customers who could actually put themselves in danger.

Waymo, GM Cruise and others all use Lidar as a 360-degree redundant sensor with cameras and radar.

We believe in redundancy also. Especially if you go fast speeds. Interviewed by Tom Krisher. Edited for clarity and length.