21–30 of 77 posts
So I'm not up on the Lidar industry but $12k for a sensor seems really expensive, but then again from my casual observations Lidar is really expensive. Is there a physical / first principles reason this is true or is it just really new technology?
The problem is that deep learning should not be allowed in safety critical systems, because (1) the accuracy is always less than 100% even in known test situations, and (2) we don't know how it works and under what conditions it breaks down.
The hardware is described, vaguely, at [1]. It's a rotating drum scanner with 16 or 64 lasers. $12,000 for the 64-laser model. $24,000 for the upcoming model with 200m range. Still a long way from a useful auto part. There are about a dozen companies in this space now. Nobody has the price down yet. Continental, the European auto parts company, is the most likely winner. Quanergy made a lot of noise but didn't ship m…
Flash lidar seems to be a fundamentally broken concept to me wrt range.
So I'm not up on the Lidar industry but $12k for a sensor seems really expensive, but then again from my casual observations Lidar is really expensive. Is there a physical / first principles reason this is true or is it just really new technology?
Image orthicons for TV cameras once cost $10,000 each, and color cameras needed three of them, plus another $50K or so of electronics to drive them. Today a cell phone camera costs about $10.
The problem is that deep learning should not be allowed in safety critical systems, because (1) the accuracy is always less than 100% even in known test situations, and (2) we don't know how it works and under what conditions it breaks down.
The status quo is neural nets already are allowed in safety critical systems - humans. And with the amount of drivers snapchatting behind the wheel, I'd rather take my chances with a self-driving car.
The hardware is described, vaguely, at [1]. It's a rotating drum scanner with 16 or 64 lasers. $12,000 for the 64-laser model. $24,000 for the upcoming model with 200m range. Still a long way from a useful auto part. There are about a dozen companies in this space now. Nobody has the price down yet. Continental, the European auto parts company, is the most likely winner. Quanergy made a lot of noise but didn't ship m…
Why Continental? Flash lidar seems to be a fundamentally broken concept to me wrt range.
There's a tradeoff between field of view and range. Automotive systems will probably include a long-range narrow field of view unit and a shorter range wide field of view unit.
Flash LIDAR has some advantages. No moving parts. Can be fabbed by semiconductor processes. The one big laser is separate from the sensor array, which helps with cooling. Also, you can spread the outgoing beam, which helps with eye safety. (Eye safety involves how much energy is in an eye iris sized, 1/4" or so, cross section of the beam. If the beam is spread out, energy density is lower.)
Because you're collecting NIR ambient light, your optics are wideband. Meaning that daylight would have a more pronounced negative effect on system range (easier to saturate the photocells). It's also low resolution (as most LIDARs are), and there is no color segmentation data.
In an automotive application, I can't see a justification to unify both visual and LIDAR into a single sensor, rather than having an extrinsically calibrated array of sensors. You can improve the calibration out of the data over time if you're very concerned about system stability.
It seems like a nice party trick, but the vehicle LIDAR game focuses on solid state long range units, as this will be what gets into mass production. The visual band imagers in the car are a given for many other reasons anyway.
Earlier quoted context omitted.
Why Continental? Flash lidar seems to be a fundamentally broken concept to me wrt range.
One might think that, except that Advanced Scientific Concepts, which Continental bought, has had it working for a decade.[1] Their units work fine, but are expensive. They're mostly sold to DoD and used for space applications. The Space-X Dragon spacecraft uses one for docking. There's a tradeoff between field of view and range. Automotive systems will probably include a long-range narrow field of view unit and a sh…
Earlier quoted context omitted.
Why Continental? Flash lidar seems to be a fundamentally broken concept to me wrt range.
One might think that, except that Advanced Scientific Concepts, which Continental bought, has had it working for a decade.[1] Their units work fine, but are expensive. They're mostly sold to DoD and used for space applications. The Space-X Dragon spacecraft uses one for docking. There's a tradeoff between field of view and range. Automotive systems will probably include a long-range narrow field of view unit and a sh…
So I'm not up on the Lidar industry but $12k for a sensor seems really expensive, but then again from my casual observations Lidar is really expensive. Is there a physical / first principles reason this is true or is it just really new technology?
And no, it's my understanding that it's just a matter of volume.