Lidar for Autonomous Cars
Optics and photonics are "hot" again.
A new generation of lidar companies supplying the self-driving car industry is putting optics and photonics back at the center of the technology world.
“Photonics is instrumental in making self-driving cars safe and ubiquitous,” says SPIE Fellow Jason Eichenholz, cofounder and CTO of Luminar Technologies, a US-based photonics startup that announced in April it is beginning production of a lidar system with an InGaAs design receiver operating at the 1550nm wavelength.
Luminar is one of the newest players in a fast-growing field of around 80 competitors such as Tesla, Google, and Quanergy Systems.
“In the late ‘90s, optics was king,” says SPIE member Louay Eldada, founder and CEO of Quanergy Systems, one of the leading companies working on the technology for autonomous vehicles. “Now it’s king again, because of lidar, the hottest area of technology.”
Both entrepreneurs have years of experience commercializing optics and photonics technologies.
Eichenholz has spent 25 years actively involved in laser, optics, and photonics R&D at Newport, Ocean Optics, Aerosonix, and Open Photonics Inc.
Eldada led the photonics unit at Telephotonics, acquired by DuPont, and was CSO at SunEdison. He also led Telecom Photonics, which was purchased by Honeywell, during the optical telecom bubble-and-bus period of the late 1990s.
He has been with Quanergy since 2012. The company now has 130 employees, with the figure likely to increase to 200 by the end of the year, at 10 locations around the world.
Among Quanergy’s 300 partners are the auto giants Mercedes-Benz and Nissan, along with parts firms Delphi and Koito, and sensor component maker Sensata.
Luminar has 150 employees after being in “stealth” mode for five years while perfecting a system touted as being able to see at 50 times greater resolution and 10 times longer range than current systems. Cofounded by SPIE member and inventor Austin Russell, it is set to begin a 10,000-unit production run of its lidar system in a Florida manufacturing facility, with deliveries later this year.
While lidar systems have become a familiar sight on early versions of self-driving cars, typically sitting rather bulkily on top of the vehicle roof, a key focus today has been to shrink both the size and cost of the technology.
During a standing-room-only keynote talk at SPIE Defense + Commercial Sensing (DCS) in April, Eldada showed a photograph of a Mercedes coupe featuring two of Quanergy’s solid-state lidar sensors.
“You don’t see the ‘lab bench’ on the roof of the vehicle,” he said. Instead, he held up a small box, the size of two playing cards, containing the core of his product. This is what “hides” behind the grille of the red Mercedes, Eldada told the crowd. That product, which can crunch through half a million data points per second to generate a live view around the vehicle, is Quanergy’s “Gen 2” effort. It costs about US$250 in large quantities.
“It’s really cheap stuff,” he said. “The magic is in the photonic circuitry.”
The third-generation version will have all the components on a single substrate, the entire lidar on one chip, he said. “It could go inside every light switch in every home,” Eldada claimed.
Eldada predicts that the cost of a lidar sensor — not long ago measured in the hundreds of thousands of dollars — will be just $100 when produced in volume in the future.
Eldada’s company has a five-fold mission, the CEO said, to save lives, as well as space, time, energy, and cost. Outside of transportation, its target sectors include security, industrial automation, and 3D mapping.
Interestingly, Eldada took issue with the “Level 1 to 5” system that the industry has adopted in order to describe the level of automation in self-driving cars.
In that breakdown, Level 1 involves minor driver assistance, while at Level 5 the car is entirely autonomous, with no human control, allowing a driver the option of sleeping in the back seat of their vehicle. Level 2 provides automated parking and cruise control to keep a set distance from traffic ahead.
Eldada sees problems at Level 3, where a driver is expected to take over control of a vehicle in certain — perhaps unpredictable and hazardous — circumstances. “Maybe industry should skip Level 3,” he suggested. “If someone, say an older person, is taking a nap, and the car says to take over now, that can be a difficult situation.”
Most carmakers want the options well defined by them, he added. “For them, at 100 times per second, they want to decide what risks they want to take for the car.”
- Have a question or comment about this article? Write to us at firstname.lastname@example.org.
- To receive a print copy of SPIE Professional, the SPIE member magazine, become an SPIE member.