MENU

Californian startup boosts LiDAR’ specs with proprietary InGaAs design

Interviews |
By Julien Happich


Gathering a team of experts and scientists which according to the company’s executives, combine over 700 years of experience in LiDAR systems, the startup now boasts that with four major autonomous vehicle programs selected as strategic partners for early testing, two acquisitions and $36M in seed funding, it will be ready to produce and ship a first pilot run of 10,000 InGaAs LiDARs this year.

The five-year old company’s boldest claim is that by using a proprietary InGaAs receiver cell operating at the eye-safe 1550nm wavelength, it is able to operate its LiDAR at a power 40x higher than silicon-based systems operating at the 905nm wavelength while remaining eye-safe.

As it likes to summarize in its specification sheets, “for every one photon a 905nm LiDAR system can safely emit, Luminar’s 1550nm LiDAR can use 68, resulting in 40 times more power, a 50 times greater resolution and 10 times longer range than current state-of-the-art systems”.

Although Jason Eichenholz, CTO and co-founder of Luminar remained elusive regarding the actual architecture of the breakthrough InGaAs LiDAR, he shared his views on the industry with eeNews Europe.

“We looked at edge-cases and focused on our customers’ needs. What customers want is a 200m range minimum with less than 10% reflectivity. When they operate their self-driving cars, they need to be able to detect hard-to-see, low-reflectivity objects like a black car or a tire on the road. There are no solutions today that provide that range at that level of reflectivity” the CTO said.

“LiDARs on the market list a detection capability at 10% reflectivity in the 30 to 40m range. At highway speed, that gives you less than 1 second of reaction time. Our system sees further out, 200m and beyond” Eichenholz added.


Discussing the technology, the CTO said: “We don’t need an array of lasers, we use a single laser system that was designed from the chip-level up specifically to address and adjust the scan pattern of our single beam. But key to seeing out 200m and beyond under 10% reflectivity was making the move from 905nm to 1550nm with 40 times the laser power”.  No MEMS-based design for beam steering either, declared the CTO without wanting to say more.

Eichenholz describes 1550nm as a very mature technology extensively used in fibre communications and for which there has been a lot of R&D, so there is know-how to reach economical scale.

“The challenge was on the receiver side, it required more effort for the overall system architecture. In order to get millions of points per second and meet our range at low-reflectivity requirements, we realized we had to make the jump to the eye-safe region. InGaAs photodiodes offer a solution where everyone else had written them off as too expensive. We only use one receiver per laser” the CTO revealed.

A roof-top mounted LiDAR unit.

Luminar’s LiDAR scans are really impressive and their spatial resolution so fine that they allow for simpler image processing, making it possible to truly determine what’s in the field of view and for example, differentiate a child from a traffic column on the side of the road.

In Q4, the company will start volume production of its next generation system with a 10,000 unit pilot run in its 50,000-square-foot manufacturing facility in Orlando. “Now and in the future, our production is fully vertically integrated in the USA” stated the CTO.

So how will Luminar compete on price or how will it position itself on the market?

“What happened over the last decades was that in order to meet the cost target of the automotive industry, the performance of LiDAR was proportionally reduced to get the volume and price down. But the price race to the bottom is the wrong race” reacted Eichenholz.

“We aim our products at level 4/5 autonomous vehicles and we aim to provide the highest quality data, so we will not race to the bottom. We are already shipping systems to AV partners and we are meeting their price requirements. They are comfortable with our trajectory for cost. Our LiDAR is already at test facilities and our partners get a level of resolution they never thought was possible”.


Coming back to the pilot run starting Q4, Eichenholz claims it will be the largest product run to date as far as LiDARs are concerned. The company will ramp up production with additional capabilities.

“One thing that is holding back the scaling of autonomous fleets is the availability of LiDARs. Luminar Technologies will be a catalyst for the growth of the autonomous vehicle industry” says the CTO.

“The quality of the data makes the image processing significantly easier. Today’s LiDARs have severe limitations in the quality and the range of their data points as well as their point density. Typically, companies throw software at it but fundamentally, it is a hardware problem” explains Eichenholz.

“We believe the solutions we are providing will be significantly interesting to AV makers. So far, it has been frustrating to see the specifications from competitors. No one ever specify all the parameters at once, reflectivity, range, point density. We deliver millions of points per second and we can dynamically adjust where the points go, zooming in an area while still viewing everywhere.

We can adjust our vertical field of view to go up or down, based on how we scan the laser beam. We can also mitigate interferences from other LiDAR-equipped cars. We’ve taken all that into account”, the CTO said.

Luminar’s prototype LiDARs embedded for test.

Revealing a bit more about the InGaAs photodiode used in the LiDAR: “We specify an epi-wafer design from a US manufacturer. We’ve figured out to best detect eye-safe laser beams and we use a silicon ASIC to do the heavy lifting to approach the price point of silicon-based solutions, but with 40 times more power and a million times more energy per pulse. We have 68 times the number of photons budget than anyone else using a silicon design”.

“We are pushing the boundaries of physics and the speed of light is a fundamental limitation. Light is too slow”, jokingly complained the CTO, “but we are not limited by what you can buy off-the-shelf, we specifically designed all the components for our purpose. We had started with off-the-shelf components but we quickly realized that this couldn’t work”.

As it will grow Luminar Technologies will offer solutions with more than one laser channel on its product roadmap, but as Eichenholz put it to conclude the interview “we are not a slideware company but a hardware company”.

Luminar – www.luminartech.com

 

Related articles:

Velodyne to produce one million LiDAR sensors a year by 2018

Quad-channel laser in scanning LIDAR prototype

Laser-pulse lidar startup wins funds

Startup raises funds for lidar development

Another move in the Lidar technology market


Share:

Linked Articles
eeNews Europe
10s