The state of LiDAR technology for the auto industry explored by Eric Volkman.
It’s a small word (technically, acronym) with only five letters. But LiDAR looms very large in the vehicle tech space. In fact, going forward it’s going to be indispensable.
Despite the advanced know-how required by the technology and the high barriers to entry in the business, there are around 50 dedicated LiDAR hardware makers just now. With the coming of full autonomy there’s a gold mine out there but it’s not particularly easy to go digging. LiDAR systems are based on complex, intricate technology built around pricey components. Still, LiDAR makers are pushing hard towards improving their systems and there’s plenty of improving to do.
“The biggest challenge LiDAR makers face at the moment is achieving the type of 3D scanning performance required for fully autonomous driving, at the safety, reliability, cost and size that is required for mass commercialisation,” said Omer Keilaf, CEO and co-founder of Israel-based LiDAR maker Innoviz. Next year the company is scheduled to launch Innoviz One, a product that it promises will be cost-effective and compact enough to integrate comfortably into a vehicle.
Stephen Crouch, CTO of Blackmore, feels that currently available technology needs improvement. “Nearly all automotive LiDAR systems are time of flight – a combination of laser pulses with sensitive detectors that ‘directly detect’ the round trip time of the pulse,” he explained. “These systems struggle to meet long range requirements (>200m) while staying eye-safe (i.e. solving the problem with more powerful laser pulses isn’t viable).”
This solution is for its system’s lasers to transmit at a steady level of optical power, as opposed to traditional pulse-based solutions. In addition to being safer, it helps produce minutely detailed maps of the terrain on the fly.
The high costs of traditional LiDAR components drove specialist manufacturer Quanergy to build its solutions around solid-state technology. This eliminates the need for a spinning laser (a feature of standard LiDAR solutions) within Quanergy sensors and reduces size and cost. Last year, the company rolled out its S3 sensor, which at a price of roughly $250 (£177) per unit was far cheaper than what carmakers and auto-driving specialists were used to paying for LiDAR.
Louay Eldada, Quanergy’s CEO, is a true believer in the future of solid state. Going forward, he says his company plans to intensify its efforts to deliver affordable and robust solutions built with such components.
Not everyone believes solid state is the glowing future of LiDAR. One of those individuals is Austin Russell, CEO and founder of another specialist firm, Luminar. In his view this technology, “is a direct response to the push for low cost LiDAR but that isn’t the solution to making self-driving cars safe and ubiquitous”. He added: “Despite a lot of efforts to reduce costs of LiDAR, there have been no significant performance advancements.”
For Luminar, the way forward is high-performance solutions, with relatively powerful sensors that can view the full world around them in 3D in real time. These “eyes” should have a range that allows for enough reaction time for the vehicle, combined with the ability to “see” in any weather condition. Luminar makes its own components that, Russell says, allow it to build sensors that have 50 times more detailed resolution than a typical LiDAR setup and with ten times the range. The issue with high-performance is, of course, the expense; these kinds of solutions are not cheap.
Yet, there is light at the end of the cost tunnel. As with any new-ish technology, LiDAR has come very far in a relatively short stretch of time. Only a few years ago a proper LiDAR solution would have set a carmaker back tens of thousands of dollars – we now live in a world where a manufacturer need only pay a fraction of that for sensors. So, gazing into our crystal ball, we can expect prices to continue heading south, while feature sets improve. It’s very likely that the current schism between low cost and high quality will narrow out into something approaching a happy medium. Which would be a happy situation for carmakers driving fast towards autonomy.