I was wondering if anyone else has experienced what I have experienced when using cruise control in my 2019 Hyundai Electric, and if they know why. So I am driving down a highway that also has stop lights. The speed limit is 80 and I'm doing 85. I'm using the smart cruise feature which causes the car to stay a certain distance behind the car in front of me and also comes to a stop when the car in front of me stops. However, if I am doing 85 km/hr and I'm coming to a stoplight where cars are sitting still waiting for the light to turn green my car does not slow down soon enough to stop safely so that I have to manually apply the brakes. Is this because the radar can't see far enough ahead to see the stopped car and slow down sooner?
Radar system discards all static (zero speed) signal echos bouncing off. As cars were stopped completely they were regarded by radar as other static items on and around the road, so they were filtered out. But as you were closing in then emergency braking system kicked based on data received from front camera system which uses machine learning image processing techniques to discover possible crash scenarios. Basically, most radar systems are bad at discovering complete traffic jams. They are relying for that on camera systems which have limited range compared to radar. So in complete traffic jam/stop scenarios never rely on smart cruise control to do its job. Apply braking manually, as for radar stopped cars are invisible and for camera you just might come too close and too fast for system to react and stop completely in a timely fashion as they are simply short-sighted.
That sounds crazy, that the radar discards static signals from objects on the road. So if a huge boulder falls onto the highway from a cliff alongside the highway the car using smart cruise would drive into it because it filtered out the static item? What is the purpose of radar if it ignores static (zero speed) obstructions in front of the car?
That is how how first generations of stealth aircraft evaded air defences. By flying low and slow. So that radars could not discriminate between clouds being bounced around by high wind speeds and an actual aircraft with low radio signal bouncing profile. The problem of radar is it its low resolution. It is very difficult to discriminate between static cars on the road and other static objects like fence. That's why radar processing logic uses the speed of objects as a discrimination what is a valid "target" and what it is not. Something which moves on the road 50 MPH most likely is not a fence but a moving car, truck or bus. Something which is static in 99,9% not a moving car. Which is correct until this is a traffic jam you will crash into. But in a same way you can crash into static fence on the road. Radar will not react to it, most automatic emergency breaking system will ignore it not to trigger unwanted breaking. All those systems, even Tesla ones, require a present driver. Real autonomous driving systems use high resolution lidars in parallel to radars and cameras to derive proper 3D image of its surroundings. That's why current Teslas will NEVER get real full self driving. It can not happen as cars are simply too blind and shortsighted with current built in sensors. They can improve the algorhytms, but you simply can not process data which is simply missing from the sensors.
Yes Smash Thus the Tesla driving into the side of the truck If the system paid attention to static objects, it would panic stop for highway signs Weakness of the system
Exactly. Even Tesla has similar deficiency. Compared to an average electric Hyundai it just has much more power-full front cameras (it basically has 3 instead of just 1) and is able to detect stopped cars better and quicker than a Hyundai. Yet it might still fail at other static obstacles like capsized truck in a similar fashion ...
