CONTACT US
Back to all articles

Sometimes You Get Lucky

Customers Are Finally Recognizing the Practical Limitations of Time-of-Flight LiDAR

When Chris and Steve started Voyant they made all sorts of decisions to get their ideas for chip-based-LiDAR out of the lab and into cars as fast as possible. Like everyone making decisions without perfect foresight, they could not anticipate the impact their decisions would have on Voyant’s future, nor the doors their decisions would open and close.

We cannot know if  Chris and Steve were lucky or smart.  Fortunately for Voyant, their decisions turned out to be the right ones.

Take Voyant’s choice of operating wavelength as an example. Voyant’s devices operate on a center wavelength of approximately 1550 nm, an invisible wavelength of infrared, referred to in technical circles as “C-band.”  Another practical choice could have been “O-band” which is more mainstream in optical datacom and offers a wider range of components.

The existence of these different optical bands has to do with quantum physics and affordable chemistry. Both C-band and O-band were intended for optical datacom, where the medium is a fiber optic strand of glass. They are not necessarily ideal choices  for LiDAR in Earth’s atmosphere.  Chris and Steve avoided the pitfall of working towards a possibly technically superior LiDAR wavelength for which there was no supply chain.  It might sound silly  but at least one well-funded startup started down that path, trying to create unique light sources and receivers from scratch for a higher performant LiDAR wavelength.

Chris and Steve did not make an exhaustive analysis of light source strategies, even between O- and C-band. They used C-band because it worked at the Lipson Nanophotonics Lab, and there was a mature supply chain for C-band sources readily available from industry

Voyant would later learn that while O-band is fine for pulsing bits down fiber-optic cables, it has serious problems traveling through the atmosphere, especially on humid days,  compared to C-band. Other companies that choose to work with O-band now face challenges in bringing their products to market.

Was choosing FMCW over  Time-of-Flight (ToF) LiDAR another example of smart or lucky Voyant engineering?

FMCW is the only practical choice for a LiDAR chip.  Optical paths on semiconductor chips are “leaky” and do not handle high-power bursts of light.  Pump too much optical power into a chip and you lose signal.  In ToF solutions, higher peak power equals better performance.  ToF solutions do not work well on Voyant’s flavor of optical chip.

FMCW LiDAR requires a higher component count than ToF, just as an FM radio has more electrical components than an AM radio. When you build from discrete components, more components means more cost. Components are free on an integrated circuit, which is why FMCW LiDAR is only practical on optical integrated circuits.

What we did not expect was how much better the FM LiDAR would perform over ToF in real applications.

One critical issue with ToF LiDAR is its susceptibility to “blooming”, which remains unresolved and might be impossible to resolve. “Blooming” is slang for distance measurement errors from very bright or reflective objects in a scene.  A LiDAR transmits its own energy to detect objects, so reflective objects like stop signs, traffic cones, or automobile lights are bright to a LiDAR even at night.

At a recent automotive LiDAR conference in Europe, LiDAR vendors walked the audience through their blooming mitigation strategies and demonstrated systems on the street.  A LiDAR can apply filters to hide blooming, wide bands or streaming tails of erroneous measurements slicing across a display, but filtering comes at the cost of not being able to see objects where filters remove data.  In real life, “blooming” means a system cannot detect the car next to the bus, or the toddler next to the traffic cone.

Another major drawback of ToF LiDAR, especially in implementations using VCSEL/SPAD (Vertical-Cavity Surface-Emitting Laser/Single Photon Avalanche Diode) architectures, is its limited range performance. Pulses can only get so big before eye safety and power become limitations.  The more power in a pulse, the more blooming.   These  limitations, where solution providers are essentially fighting physics, become particularly pronounced in automotive markets where the demands for range and precision are highest. Consequently, ToF solutions are likely at a massive disadvantage compared to low-cost FMCW solutions.

In contrast, FMCW LiDAR is rapidly gaining recognition as the future of LiDAR technology. OEMs and Tier-1 are actively searching for FMCW LiDAR solutions, promising deployments within the next 3-4 years.

Steve and Chris were already committed to FMCW LiDAR before Voyant learned how inadequate ToF solutions might be, and before Voyant had fielded systems to learn just how good Voyant’s products were in the real world.  Another lucky break for Voyant?

The biggest “Luck over Perfect Knowledge” break for Voyant is around cost.  There are plenty of ToF solutions in the market, and even an FMCW LiDAR.  For large scale market penetration, LiDAR prices have to fall by a substantial amount.  Whether this price reduction is a factor of 5 or a factor of 50 depends on your perspective, but the lack of significant design wins in any market from any company is a strong indication that no current product is offering a clear winner on price and performance.

It’s generally acknowledged that if you can mass produce a solution on a small semiconductor chip it will beat a solution made from macroscopic bits of stuff.  Generally.  But Chris and Steve didn’t start out with a complete system design or cost model, and could not be sure how big their on-chip designs would be, and thus did not know how expensive their LiDAR chips would be to manufacture.  Surface area, process costs, and – most importantly – chip yield, were all essentially unknowns.

We will never know if the Voyant team was lucky or smart.

What we do know is that Voyant offers industry leading cost and performance for LiDAR.

Please come find us at CES, and we will show you!

Peter Stern

AUTHOR

Share This Post

Machine perception is within reach

Our APIs get you the most from our sensors without getting stuck in the photonic weeds. Leave the details of LiDAR to us, so you can focus on your application.

CONTACT US