AIR.LUCID
Active Member
- Joined
- Nov 22, 2021
- Messages
- 479
- Reaction score
- 212
- Location
- El Dorado Hills, CA
- Cars
- ‘04 Prius, ‘98 MB SLK230
Does anyone know when LIDAR will be implemented in DreamDrive Pro?
Interesting, apologies, BMW was claiming to use some kind of laser ranging at least 10 years ago.I remember it as LIDAR not RADAR. I had never heard of LIDAR before I reviewed and bought that car, and it impressed me. But, hey, it was a long time ago.
No apologies needed. There is so much technological advances going on and so many claims being made by so many sources that it is hard to keep the facts from the fiction. Best regards.Interesting, apologies, BMW was claiming to use some kind of laser ranging at least 10 years ago.
Just watch the Video I posted. They have even broken these down by type of obstacle on their Website and LinkedIn and Twitter.I don’t think it would be able to see multiple cars ahead unless it has line of sight. If there is a car that is concealed by a truck in front of you, LiDAR won’t “see” the car the same way a camera or human eye wouldn’t.
Radar can often see the car in front of a truck in front of you. At least I found that in my Model 3 frequently.I don’t think it would be able to see multiple cars ahead unless it has line of sight. If there is a car that is concealed by a truck in front of you, LiDAR won’t “see” the car the same way a camera or human eye wouldn’t.
You might want to read up on how radar works.Radar can often see the car in front of a truck in front of you. At least I found that in my Model 3 frequently.
That was my theory. It happened often.Really!? That is amazing! I wonder how that works? Maybe some of the radar waves bounce under the truck, hit the car and then bounce back under the truck to your sensor? You must have a sensor with very high sensitivity. How cool is that!
I'm a career electrical engineer of nearly 50 years' experience, with a great deal of time spent designing and embedding sensor systems into electromechanical products. I think what I saw in my Model 3 was due to ground bounce of the radar signal under the truck. Just saying what I observed - and was surprised by it!You might want to read up on how radar works.
I did not know radar waves went around corners and were in fact only accurate in direct line of sight. Sure they could bounce off the ground but the the chance of them being reflected at the same angle back underneath a truck in front of you is fantasy.I'm a career electrical engineer of nearly 50 years' experience, with a great deal of time spent designing and embedding sensor systems into electromechanical products. I think what I saw in my Model 3 was due to ground bounce of the radar signal under the truck.
I'd disagree - reflections from both the ground and the car in front would be scattered so some energy would make it back to the receiving array - but the signal would be quite attenuated. It's the only simple explanation for what I observed many, many times with Tesla's simpler Autopilot software stack. Their Full Self Driving software I used for a short while could also often keep track of cars in front of trucks, but this was only optical, with some kind of object modeling that included a degree of object persistence and trajectory prediction when direct view was occluded.I did not know radar waves went around corners and were in fact only accurate in direct line of sight. Sure they could bounce off the ground but the the chance of them being reflected at the same angle back underneath a truck in front of you is fantasy.
How is this going to work accurately when you have hundreds of cars per mile on a two way road sending out waves that are reflecting all over the place. How will one car differentiate what it is receiving back in information when the air is inundated with hundreds of sending units?I'd disagree - reflections from both the ground and the car in front would be scattered so some energy would make it back to the receiving array - but the signal would be quite attenuated. It's the only simple explanation for what I observed many, many times with Tesla's simpler Autopilot software stack. Their Full Self Driving software I used for a short while could also often keep track of cars in front of trucks, but this was only optical, with some kind of object modeling that included a degree of object persistence and trajectory prediction when direct view was occluded.
Sensing in a noisy environment generally requires several mitigation techniques used simultaneously. For pulsed-excitation sensors, like Lidar, some Radar and Sonar, a common method is to modulate the transmitted signal so that when an echo is received, you can use correlation methods to tell it apart from noise and other transmitters. Like yelling your name at a cliff face and waiting for the echo of your name. Other methods include transmitting in very short pulses, spaced far apart and randomly, and only accepting return signals occurring within a reasonable time after the transmit pulse. Or transmitting a burst on a specific frequency, and changing the frequency with every pulse (frequency-hopping). After all this, there are various computational methods used to filter the received signal - at the simplest example, you could only accept a reflection if it occurred for several consecutive samples.How is this going to work accurately when you have hundreds of cars per mile on a two way road sending out waves that are reflecting all over the place. How will one car differentiate what it is receiving back in information when the air is inundated with hundreds of sending units?
Try it again with Adaptive Cruise control. When active, this brings me to a complete stop when approaching a stopped car at an intersection for example. Then when traffic clears, a upclick on the left steering wheel toggle resumes forward motion at the preset speed.Yes, my early collision system is on, but when approaching stopped traffic at over 35mph, I would rear end the car in front of me, if I wait until the brake warning for it to stop. The system also does not ‘see’ traffic if there is a bend in the road. There are too many edge cases to enumerate here. Let’s just say it works much worst than the Tesla system, which has no LIDAR.
This is my experience too. ACC has been *spectacular* at detecting stopped traffic. Better than any system I’ve used before.Try it again with Adaptive Cruise control. When active, this brings me to a complete stop when approaching a stopped car at an intersection for example. Then when traffic clears, a upclick on the left steering wheel toggle resumes forward motion at the preset speed.
That is not a problem for me as I will pass away long before I will ever trust any car to park for me.This is my experience too. ACC has been *spectacular* at detecting stopped traffic. Better than any system I’ve used before.
Auto park, on the other hand, is a bag of crap.
Actually I also have the Adaptive Cruise Control on and about 40% of the time the car will not stop if going over 35mph. What I usually do now is when approaching stopped traffic quickly I "feel" to see if it will slow down. If not, I disable ACC and when I am closer and going slower I reenable it. In any case, this system is not ready for prime time yet. Maybe when they bring on the Highway Assist mode, the LIDAR will be tuned better.Try it again with Adaptive Cruise control. When active, this brings me to a complete stop when approaching a stopped car at an intersection for example. Then when traffic clears, a upclick on the left steering wheel toggle resumes forward motion at the preset speed.
We must be driving in different conditions, because I always cringe when approaching stopped traffic. The Tesla system, by the way, has behaved almost flawlessly.This is my experience too. ACC has been *spectacular* at detecting stopped traffic. Better than any system I’ve used before.
I wonder if this is an issue with your sensors? I’ve tested the same scenario and it did brake perfectly. There have been one or two times that it warned me to brake instead, but overall it has handled stopped or slowed traffic spectacularly.We must be driving in different conditions, because I always cringe when approaching stopped traffic. The Tesla system, by the way, has behaved almost flawlessly.
One example is that I drive on a road where the speed limit is 45mph. I drive with traffic which is usually going about 49mph. There is a left hand turn in a median that I take. Many times there are cars already waiting to turn left. If I steer into the left hand turn lane, the car will not brake if the cars ahead are not moving. I also need to be fully in the lefthand lane before it will even consider braking. In the Tesla, I can steer into the lefthand lane at full speed and the car will slow down, even before I'm fully in the lane.
Other examples are the typical car in front of you quickly changing lanes to reveal a stopped vehicle. The Lucid will try to plow into the other vehicle if I'm going over 35mph. The Tesla vehicle will slow down and stop appropriately.
I can go on with various examples, which I think indicate that the LADIR, if used, is only trained on the vehicle ahead (if at all) and not aware of it's surroundings. That along with the system not intelligently assessing the curvature of the road or road markings, makes me think that it has a long way to go.
My sensors seem to be like @Alex also. I will test it out more today. Other than stopped vehicles the ACC is excellent.I wonder if this is an issue with your sensors? I’ve tested the same scenario and it did brake perfectly. There have been one or two times that it warned me to brake instead, but overall it has handled stopped or slowed traffic spectacularly.