Or listen in your favorite podcast app
Autonomous cars aren’t the future, they are the present. And as more and more companies invest in the world of autonomous transportation, a market for a very specific kind of tech has grown bigger as well. Jun Pei (LinkedIn) is the CEO of Cepton, which makes Lidar systems. Lidar is an essential piece of technology that measures distance at a rate that is accurate down to the millimeter. So, if your autonomous car is driving along and an object flies in front of you, the Lidar system can not only detect it but also measure how far away it is and the rate at which you are moving toward each other.
On this episode of IT Visionaries, Jun explains all of that and more, and he dives into how Lidar will become even more necessary in the near future.
Best Advice: “Stay calm. You’re going to experience a lot of issues you haven’t experienced before, so stay calm and solve them one by one.”
How Jun got interested in technology — (1:45)
At a young age, Jun saw how a prism split sunlight into a rainbow and ever since then he wanted to study optics and technology. He followed his passion to Stanford and has stayed in Silicon Valley ever since. He says that as a Stanford graduate, there is a certain expectation that you will go on to start your own company or come up with a breakthrough technology. That’s both a motivator and something that weighs on you.
“I certainly felt the pull….Plenty of friends and classmates started their own companies. Some of them made it really big, and there’s always an urge that if you’re graduating from Stanford especially in the electrical engineering department, you need to do something, start a company and make a difference.”
The Cepton story — (3:35)
Cepton started three years ago and makes Lidar. Lidar wasn’t known to everyone until just recently. It used to be a very high-tech exclusive technology but it has recently become a default tech necessary for autonomous vehicles. But, when companies first started investing in autonomous vehicle production, there were only a few Lidar companies in the world. Jun and his co-founders saw a gap in the industry to create high-quality low-cost products, especially because Lidar, Jun says, are the true eyes of the vehicle.
The name Cepton is a made-up word. The “Cep” comes from “perception,” and “ton” from an atom. So it’s an “element of perception.”
“There’s a very natural analogy to a human being that the vehicle, the drive train itself, the wheels, and the engines are like your legs and lidar is like your eyes. And then the computing system is like your brains. So when you talk about a human being, you’re going to talk about their eyes and legs, but you are going to talk a whole lot more about their brains. But on the other hand, you can understand how important eyes are for human beings, and so this is by far the most important sensor for an autonomous vehicle.”
The future of Lidar — (9:10)
Today’s cars are equipped with all kinds of sensors and cameras designed to make driving safer. Plus, they offer ways to keep you up to date on the health of every part of your car, which heads off problems before they become major headaches. But, when it comes to cameras specifically, they have no sensing technology to tell you the distance or movement of what it’s seeing. Lidar gives you that capability and allows you to measure movement and velocity. Lidar also provides a high level of confidence for a number of scenarios from mapping to security and so many other situations.
The applications of Lidar still come at a high cost in a lot of industries. Jun points to how drones have become more prevalent and with that development, the need for Lidar could increase leading to an opportunity in the market. Because drones demand a requirement for a lower cost, lightweight lidar devices are the perfect tool. Jun also explains that Lidar can be used for topology, 3-D mapping and even disaster efforts. So whether that’s wildfires, earthquakes or whatever, Lidar can help accurately survey an area.
“Without any human intervention, our software can easily determine whether an object is of any harm or not. This is a third dimension. A camera measures the x, y, but it doesn’t know the distance. So this third dimension, the distance information, is indeed what provides you with a high level of confidence in many scenarios whether it’s for security or for mapping or for any other applications.”
The pros of Lidar — (20:30)
One of the best uses of Lidar is to accurately measure distance. And while Lidar can be incredibly accurate, with autonomous cars you actually need less accuracy and more reliability. However, less accuracy doesn’t mean you’re sacrificing safety, it just means that you’re going to be measuring in terms of centimeters rather than millimeters.
So with that said, the future is not about improving accuracy, it’s more about cost, reliability, and deployment in applications.
In the next decade or two Lidar will be just as common as cameras. The third dimension gives you an extra piece of data that’s critical while also removing a concern. Jun explains that there are more concerns with privacy when dealing with cameras. Lidar doesn’t have that issue because it doesn’t worry about facial recognition or color. It doesn’t measure the privacy-related data that people have issues with.
“Once you know the distance between you and the car in front of you to an inch, that’s enough. It’s beyond what a human eye can differentiate.”
Downsides of Lidar — (23:00)
Lidar is an active device that relies on lasers and has its own illumination source whereas cameras rely on illumination from the sun or a flash. That’s a good thing, but at the same time, it creates a problem because if you have too many Lidars together, they can actually interfere with each other.
Being a leader while building a very technical device — (26:00)
Jun was trained as an engineer and as a scientist, so working on the technical side of things was second nature for him. But there were areas he was never trained in, and he says in those areas you just have to force yourself into it. What he’s learned by doing that is that it all comes down to building the right team, and that’s what he spends most of his time working on.
“I don’t think I’ll ever be prepared enough for the role but I think I can grow into it.”
“Once you have the correct team in place, I can render myself useless.”
Building hardware — (28:40)
When it comes to building hardware, it’s a much more complex and difficult process than building software. Every single element in the system has to work before you can call it a product. That’s a challenge, but Jun actually sees that as part of the fun of the job as well. In fact, when he’s bringing new people into work with the team, he makes sure that they consider work not as work, but as a hobby that they enjoy doing.
Building hardware also means that there are certification processes that you have to go through, especially when it comes to vehicles and transportation. At Cepton, they have established a fairly rigorous process from the design phase onward to ensure that they are meeting those certification needs. Having those processes is a deviation from the path for most start-ups, but it’s what makes Cepton work. It might take more time and cause a bit more hassle, but it’s the only way to arrive at a highly-reliable, certified safe product.
“We really have to be fairly rigorous and process-oriented from the get-go. This actually is a big deviation from the concept of a startup, especially a silicon valley’s startup which is a lot of just go shoot from the hip and get things down quickly….Lidar is considered a safety device, just like your tires and brakes, et cetera. So we actually established a fairly rigorous process even from the design phase, before we even have the first product or even the first concept. We go through what’s the approach we’re going to take to actually arrive at a highly reliable product.”
Where is the market now? — (32:25)
In the automobile industry, there is a recognition and a general consensus that Lidar is a necessary part for Level 3 autonomous vehicles. There is also a consensus that it’s needed for Level 2.5, with automatic braking and such. Even though cameras and radar can take on 99.9 percent of cases of autonomous vehicles, Lidar is used for that .1%. Jun puts it this way, if one out of 1,000 drives you get into an accident, you wouldn’t want to do that and no company would create a car with that possibility. Lidar takes away that .1% of danger and so there will be no autonomous cars without Lidar.
“It’s the difference between having an autonomous vehicle or not having one. I think you’re going to see Lidar being deployed in the next three to five years and you’ll see it added to all vehicles in the future.”