Lucid's LiDAR supplier

[Adnillien]

I was finally able to dig up who is supplying Lucid LiDAR modules. It is Robosense which is using a 905nm technology. This appears to be the most common for ADAS. Robosense is China based and also supplying Xpeng, Zeeker and GAC. The lasers are EEL (Edge Emitting Lasers) made by OSRAM and the sensor/reciever is SiPM (Si Photo Multiplier) made by OnSemi. It appears to sell for around $1900. It cover 120 degrees horizontally, 25 degrees vertically and draws 15W of power.

RoboSense | Safer world, Smarter life

RoboSense (HKEX: 02498) is a pioneering robotics platform company for Physical AI. Based on its in-house optoelectronic chips and perception algorithms, and supported by its scaled mass production capabilities, RoboSense provides robots with core components and infrastructure, including LiDAR...

www.robosense.ai

 

[hydbob]

I was finally able to dig up who is supplying Lucid LiDAR modules. It is Robosense which is using a 905nm technology. This appears to be the most common for ADAS. Robosense is China based and also supplying Xpeng, Zeeker and GAC. The lasers are EEL (Edge Emitting Lasers) made by OSRAM and the sensor/reciever is SiPM (Si Photo Multiplier) made by OnSemi. It appears to sell for around $1900. It cover 120 degrees horizontally, 25 degrees vertically and draws 15W of power.

RoboSense | Safer world, Smarter life

RoboSense (HKEX: 02498) is a pioneering robotics platform company for Physical AI. Based on its in-house optoelectronic chips and perception algorithms, and supported by its scaled mass production capabilities, RoboSense provides robots with core components and infrastructure, including LiDAR...

www.robosense.ai

That's great! Is this good or bad for us owners?

 

[borski]

I was finally able to dig up who is supplying Lucid LiDAR modules. It is Robosense which is using a 905nm technology. This appears to be the most common for ADAS. Robosense is China based and also supplying Xpeng, Zeeker and GAC. The lasers are EEL (Edge Emitting Lasers) made by OSRAM and the sensor/reciever is SiPM (Si Photo Multiplier) made by OnSemi. It appears to sell for around $1900. It cover 120 degrees horizontally, 25 degrees vertically and draws 15W of power.

RoboSense | Safer world, Smarter life

RoboSense (HKEX: 02498) is a pioneering robotics platform company for Physical AI. Based on its in-house optoelectronic chips and perception algorithms, and supported by its scaled mass production capabilities, RoboSense provides robots with core components and infrastructure, including LiDAR...

www.robosense.ai

Given that that single component is $1900, the $9k price for DD Pro seems a little less outlandish, assuming other sensors and hardware are also included.

 

[Adnillien]

That's great! Is this good or bad for us owners?

I don't think it is either good or bad. Robosense seems to be picking up design wins faster than others. They are also partnering with Nvidia on the processing side which matches with the Lucid hardware.

 

[Adnillien]

Given that that single component is $1900, the $9k price for DD Pro seems a little less outlandish, assuming other sensors and hardware are also included.

I agree, it seems about right given the hardware cost.

 

[DeaneG]

Interesting. From the looks of the video footage, cameras could still be helpful to augment the lidar point cloud at freeway speeds. The cloud starts to thin out considerably past about 50 meters.
I believe lidar is necessary for safety with the state of today's computing technology, but it is amazing what Tesla has done with cameras alone so far.

 

[Tesla2.0]

Sample application

 

[borski]

Interesting. From the looks of the video footage, cameras could still be helpful to augment the lidar point cloud at freeway speeds. The cloud starts to thin out considerably past about 50 meters.
I believe lidar is necessary for safety with the state of today's computing technology, but it is amazing what Tesla has done with cameras alone so far.

Meh. If you want to see stellar video-only autopilot, check out Comma: https://comma.ai/

Is it great? No, it's meh (for now). But it is basically why Musk went all vision, after geohot challenged him. https://electrek.co/2015/12/16/elon...mobileye-crushing-autopilot-system-for-tesla/

 

[DeaneG]

Meh. If you want to see stellar video-only autopilot, check out Comma: https://comma.ai/

Is it great? No, it's meh (for now). But it is basically why Musk went all vision, after geohot challenged him. https://electrek.co/2015/12/16/elon...mobileye-crushing-autopilot-system-for-tesla/

I was referring to Tesla's "Full Self Driving beta" software, which does a nice job of discerning cars, people, etc at a distance longer than I'd expect given the resolution of its cameras. IMO a big limitation of this approach is the apparent need for its AI to recognize every possible foreign object on a road from any angle (ladder, piece of paper, pillow, sofa, tire shred) vs LIDAR noting that there is an object in the road and allowing the cat to avoid it. Haven't seen evidence of Tesla's "beta" noting stationary objects in the road other than preprogrammed ones (traffic cones, garbage cans).

 

[Tesla2.0]

Here is what Tesla camera’s limitation and owners will accuse as Phantom Brake. Lucid’s LiDAR will just drive over these.

 

[daydreamer223]

Here is what Tesla camera’s limitation and owners will accuse as Phantom Brake. Lucid’s LiDAR will just drive over these.

View attachment 4356View attachment 4357View attachment 4359View attachment 4361View attachment 4358View attachment 4360

To be fair some of these early ones broke my mind at first- great art!

 

[Sam6062]

I don't think it is either good or bad. Robosense seems to be picking up design wins faster than others. They are also partnering with Nvidia on the processing side which matches with the Lucid hardware.

That is the Industry Buzz word today. “Design Wins”. No one has a production Contract today. Two Companies Luminar and Innoviz have been making the most noise and claim Billions of Dollars worth of Wins. The latter claims they will be supplying all the VW brands with their yet to be tested version in 2025.

The 905nm laser is the most Cost effective proposition but has a higher level of danger to the human eye and need to be deployed carefully and the 1055 is less harmful but extremely expensive.

The ultimate Cost to OEM’s to fully embrace LiDar is between 500-800 per module and each Car will need at least 2.

What Lucid has is most likely limited to ADAS Level-2 right now with its single unit and augments the Cameras when active.

Has Lucid confirmed that the LiDar in the car is functional with the current capabilities of DD Pro or reserved for future enhancements?

As explained by those photos the Cameras get confused by so many things including Sunlight and even the Moon. There is a Vidoe out there where a low moon is mistaken as a Yellow Traffic light by the Tesla Camera.

LiDar would be good in any weather and any light. A perfect autonomous driving system will use a combination of LiDar, Cameras and Radar to be safe and effective at various distances, short, medium and long.

 

[hmp10]

What Lucid has is most likely limited to ADAS Level-2 right now with its single unit and augments the Cameras when active.

I don't remember the source and thus can't vouch for its credibility, but a recent video reviewer said that right now Lucid is using lidar only to feed data into their mapping database. It is not currently linked to any of the operating functions of the car itself.

 

[Tesla2.0]

I don't remember the source and thus can't vouch for its credibility, but a recent video reviewer said that right now Lucid is using lidar only to feed data into their mapping database. It is not currently linked to any of the operating functions of the car itself.

Machine learning coding? Maybe we are just uploading what we see and how we drive to respond what we see, the data gets uploaded to server to AI to learn and then deploy AI later? I’m cooled being tracked to help. Maybe that explain some Lucid bandwidth issue.

 

[Sam6062]

I don't remember the source and thus can't vouch for its credibility, but a recent video reviewer said that right now Lucid is using lidar only to feed data into their mapping database. It is not currently linked to any of the operating functions of the car itself.

I would believe that statement. I think that may have been something they added for a future enhancement when Nvidia can supply them with some really useful ADAS Tech. Nvidia themselves is just getting started in this space along with Qualcomm.

I am a biased investor in MVIS MicroVision who have also supplied MEMS Tech to Microsoft for their 20 Billion DOD contract for Hololens2 and IVAS systems to the Army that is about to be funded in October. It’s late due to the level of testing being done and the scrutiny it has due to the heavy cost. MicroVision video I posted is the unfiltered real life point cloud information. LiDar will supply that to the OEM’s domain controller to then make the final decision. LiDar will only say Drivable or Not Drivable and then it is upto the Domain controller to decide what action will be taken using the Data. Interesting stuff coming up in the next 3 Years. The tech is just now starting to come into its own and will be on display in many Cars truly using its full functionality in 2015. I know there is LiDar being deployed in some fashion in some high end cars but it is very rudimentary. MicroVision tested theirs at 130KMPH. Their LiDar is 3 in 1 short medium and long range and the OEM can pick and choose what they want. Software will do the rest.

 

[Adnillien]

The 905nm laser is the most Cost effective proposition but has a higher level of danger to the human eye and need to be deployed carefully and the 1055 is less harmful but extremely expensive.

I believe you meant 1550nm instead of 1055. You are right about the cost of the 1550nm solutions, I think the big advantage of the longer wavelength is range but that advantage is going away with improvement in the SiPM detectors at 905nm.

Here is what I have found on 2021 revenue by ADAS suppliers. The market is still very small so all of the suppliers talk of design wins rather than actual sales.
Valeo - 28.5M
Robosense - 3.8M
Livox - 2.4M
Continental - 1.6M
Denso - 0.9M
Ibeo - 0.8M

 

[Sam6062]

I believe you meant 1550nm instead of 1055. You are right about the cost of the 1550nm solutions, I think the big advantage of the longer wavelength is range but that advantage is going away with improvement in the SiPM detectors at 905nm.

Here is what I have found on 2021 revenue by ADAS suppliers. The market is still very small so all of the suppliers talk of design wins rather than actual sales.
Valeo - 28.5M
Robosense - 3.8M
Livox - 2.4M
Continental - 1.6M
Denso - 0.9M
Ibeo - 0.8M

Actually Valeo with help from IBEO is about the only one who has supplied any units that are in actual Production vehicles from Audi.

The others are not real LiDar suppliers to automotive.

They are Innoviz, Luminar and MicroVision(new to LiDar and just bought the usable parts of bankrupt German firm IBEO) MVIS is the symbol.

Innoviz and Luminar claim design wins which are not actual contracts although Volvo EX90 will have an ugly bump on the roof line with Luminar’s LiDar.

 

[Slam08]

Innoviz and Luminar claim design wins which are not actual contracts although Volvo EX90 will have an ugly bump on the roof line with Luminar’s LiDar.

Design is funny. I bet that when the first real level 5 autonomous cars come out with large LiDAR bumps it will become a cool status symbol. Because that means that the car is SELF DRIVING! That lucky person is prolly asleep or typing an email!

I could even see companies faking bumps to make their cars look more autonomous.