-
Want to remove ads? Register an account and login to see fewer ads, and become a Supporting Member to remove almost all ads.
Hello, rear radar :-)
- Thread starter lunitiks
- Start date
Dual band
I think it's for the "backwards FSD" easter egg where FSD will be released on April 1st but only in the reverse gear.
Reference for the screen-shotted FCC docs/images: FCC ID 2AEIM-1089775 Fascia Endpoint by Tesla Motors, Inc
All FCC wireless applications from Tesla: Tesla Motors, Inc FCC ID Applications (2AEIM)
Maybe we should rename the thread: "Goodby, rear radar ;-("
Side note, I guess you're happy now BigD0g, but I am still doubtful that even high speed Level 3 will work on the German Autobahn, without a rear radar. And IMO even merging into a highway would need a rear facing radar, to get the speeds of the approaching cars and trucks right. But maybe this can be done with the cameras as well...
Oh, I think rear radar is needed, I also think front corner radars are needed, but clearly I don’t know anything
My (possibly too simple view), if the amount of road surface between you and another car is constant or increasing and sufficient, you are ok to merge. Quantitative distance and speed are needed for the decreasing case, but even then, rough approximations are sufficient. Like with a backup camera, perspective gives a decent estimate of range.
For a running merge go as fast as the car you are merging behind and verify you are not trying to merge into another vehicle.
If this is a merge from a stop, I'm not sure radar would have the resolution needed to pick out a gap in oncoming traffic, and it would be pointed non-optimally (esp if there is a concrete barrier).
US centric numbers: 60 MPH is 88 ft/sec, with a 60 Hz camera rate that is 1.5 ft per frame. Typically, if a vehicle does not also have flashing lights, you an not going to see that high of a closure rate. Perceptive dictates the position of the vehicle in the camera field of view, like the lines in a backup camera screen. Can also glean information based on rate of size change, but that is more dependent on vehicle type/ edge detection/ aspect to target.
If radar were needed, it would likely take one for each side to reduce shadowing from a following vehicle. The further out you want the radar to see, the lower the return strength you need to act on (return strength falls off to the fourth power). That increases the chance of registering noise (or a multipath return) as a vehicle/ object of interest.
Completely agreed, especially with the latter point. Rear radar like blind spot radar is notoriously ghosty because it can bounce off stationary objects (which increases the distance), or bounce off moving objects which results in shadowing and combining, which can add or subtract both distance and speed.
I'm not convinced that corner radars make merging clearly easier compared to a camera pointed in the right direction. The camera is still going to have to provide at least a distance estimate (if not a speed estimate as well) to determine whether it's correlating the visual observation with the correct radar signature. Furthermore, around curves, radar readings are already problematic in traditional radar based blind spot monitors — I frequently get blind spot falses in my Audi for cars that are 2 lanes over, especially when the road curves.
jerjozwik
Active Member
so your saying i should start fancy parking so that i can approch my car from its butt.
...for the phonekey, of course.
Zarwin
Member
Yeah, I'm wondering the same thing. Triangulation with the B pillar points? Doesn't seem like it's just for coverage since I'm inside my house right now with the Model 3 out in the driveway and still have connection to all 4 radios. If I go to the far corner of my house on the second floor opposite of where the driveway is I still have connection to 3 radios. They have some range.
With that kind of range having multiple may be a necessity to get any sort of idea where the phone is. They could also be using diffetential time of flight from the phone to the receivers for location (that may be overkill though).
ValueAnalyst
Closed
dc_h
Active Member
I’m not going to look this up, but early in the FSD discussion car to car communications was considered critical. What is the device to device range? In traffic or in a group of AP 2.5 cars, would platooning be easier if the cars could exchange metadata. Can the neural network be extended by tying the cars together locally? Avoiding cell networks seems much lower latency and cost, but limited to very close range.
Yes, that's what I meant. Regular code today, but probably some NNs+lots of custom logic when true EAP/FSD features get deployed (errr, IF they ever get deployed
).regular code is the same thing as "custom logic", so I don't really get your point then.Yes, that's what I meant. Regular code today, but probably some NNs+lots of custom logic when true EAP/FSD features get deployed (errr, IF they ever get deployed
@lunitiks good find on the ble antenna.
Now that you mention it. Its located in the exact place I would expect a standard rear RF smart fob antenna to be fitted.
20/20 hindsight most manufactures will use two radar units on the outboard edges of the rear body, to ensure as little blindspot as posible.
Now that you mention it. Its located in the exact place I would expect a standard rear RF smart fob antenna to be fitted.
20/20 hindsight most manufactures will use two radar units on the outboard edges of the rear body, to ensure as little blindspot as posible.
Similar threads
- Article
