Auto learn locking out falses with less than one pass? Actually zero passes.

timark251

PSL +5
Intermediate User
Joined
Aug 16, 2017
Messages
502
Reaction score
832
Location
Wayne, IL
So, I decided to turn Auto Learn on to help deal with the few falses I've come across with my new 360c. Odd thing happened tonight though, had a long range alert of a known speed sign, alerted for a few seconds, went away and then alerted again to the same source. After that, I got the single beep and Stored on my detector. Unlocked, alerted, 15 seconds later it alerts again and locks out the signal. Is this normal behavior for all Escort detectors with auto learn enabled? Do the detector count each time it sees it as a "pass" or am I missing someting on how it works?
 

Brainstorm69

TXCTG - 2016 MOTY
Advanced User
Lifetime Premium Member
Joined
May 23, 2015
Messages
10,333
Reaction score
23,508
Location
Lone Star State
So, I decided to turn Auto Learn on to help deal with the few falses I've come across with my new 360c. Odd thing happened tonight though, had a long range alert of a known speed sign, alerted for a few seconds, went away and then alerted again to the same source. After that, I got the single beep and Stored on my detector. Unlocked, alerted, 15 seconds later it alerts again and locks out the signal. Is this normal behavior for all Escort detectors with auto learn enabled? Do the detector count each time it sees it as a "pass" or am I missing someting on how it works?
Sometimes Autolearn just seems to be a mystery....it would be nice if Escort clearly described how the current implementation works.
 

GotWake

Speed Kills
Advanced User
Joined
Jun 4, 2011
Messages
9,695
Reaction score
11,134
Location
NC
Since many of us have seen legit k band on both sides of the segment, I can't imagine how Escort would be able to lockout a source on one pass.

I will say that my STIR+ locked out a leo I was following one time years ago. I was following this sheriff in Va and he had probably one of the lowest k band frequencies I've seen at down around 25.050.
 

timark251

PSL +5
Intermediate User
Joined
Aug 16, 2017
Messages
502
Reaction score
832
Location
Wayne, IL
Since many of us have seen legit k band on both sides of the segment, I can't imagine how Escort would be able to lockout a source on one pass.

I will say that my STIR+ locked out a leo I was following one time years ago. I was following this sheriff in Va and he had probably one of the lowest k band frequencies I've seen at down around 25.050.
I just find it odd that such a strong source, being "seen" multiple times with a line of sight, would be locked out before even passing the source.
 

thebravo

Advanced User
Lifetime Premium Member
Joined
Dec 31, 2016
Messages
6,172
Reaction score
12,741
Location
FL (formerly CT)
I wonder if the detector while autolearn is turned off still logs all the detections/ frequencies/locations but just doesn't lock them out. Then when autolearn is turned on it already has data from multiple passes so when you approach a known false it applies the data it already collected while autolearn was off and instantly locks it out. That would be kinda cool if that was indeed what they did. You then turning on autolearn doesn't then require a ton of extra passes to start locking them out... May not at all be the case here but it's possible I suppose.
 

sdrawkcaB

Defender of Light 📶
Advanced User
Lifetime Premium Member
Joined
Mar 6, 2018
Messages
5,237
Reaction score
10,324
Location
Here
Paging @poolmon , he’s a wealth of knowledge on Escort products.
 

GoWFO

If you have to go... GoWFO!
Advanced User
Lifetime Premium Member
Joined
Jul 21, 2016
Messages
1,204
Reaction score
1,263
Age
53
Location
Centre of the Canadian Universe
I wonder if the detector while autolearn is turned off still logs all the detections/ frequencies/locations but just doesn't lock them out. Then when autolearn is turned on it already has data from multiple passes so when you approach a known false it applies the data it already collected while autolearn was off and instantly locks it out. That would be kinda cool if that was indeed what they did. You then turning on autolearn doesn't then require a ton of extra passes to start locking them out... May not at all be the case here but it's possible I suppose.
That would have been my guess too
 

poolmon

Advanced User
Lifetime Premium Member
Joined
Oct 26, 2013
Messages
762
Reaction score
1,209
Paging @poolmon , he’s a wealth of knowledge on Escort products.
My understanding is that when AutoLearn is Off there will be no further logging of new data points, but that AutoUnlearn will still be active to delete already-logged data that may be going stale. ER can jump in to clarify/correct.

A lot about the inner workings of AutoLean has not been disclosed for obvious reasons but I did run across some interesting nuanced details of how AutoLearn can continue to learn in situations where you have deliberately suppressed certain alerts. For example, using the Mode Selection button to suppress an alert (such as using Auto, Auto No X, or Auto Low K ) verses suppressing an alert via Band Toggle (on/off) in Preferences will affect AutoLearn's ability to continue log data. If interested see posts 4 thru 8 at link below for how this works.

 
Last edited:

CobawLT2010

Batteries can’t save us!
Advanced User
Lifetime Premium Member
Joined
Oct 26, 2011
Messages
7,614
Reaction score
4,204
Location
Mining in the DRC!
TBH I wish escort live transmitted an auto learn database to the detector based on what live has seen in the past and other users lockout passes... if that was a feature then it’d be the bomb...
 

poolmon

Advanced User
Lifetime Premium Member
Joined
Oct 26, 2013
Messages
762
Reaction score
1,209
So, I decided to turn Auto Learn on to help deal with the few falses I've come across with my new 360c. Odd thing happened tonight though, had a long range alert of a known speed sign, alerted for a few seconds, went away and then alerted again to the same source. After that, I got the single beep and Stored on my detector. Unlocked, alerted, 15 seconds later it alerts again and locks out the signal. Is this normal behavior for all Escort detectors with auto learn enabled? Do the detector count each time it sees it as a "pass" or am I missing someting on how it works?
As I understand it, your RD reported both that it stored and then that it unlocked the speed sign?
If it reported an unlocking, this cannot be done automatically (vs a an AutoUnlearn which is silent). An unlock can only be done by a press of the Mute Button in the presence of an active signal (other than Ka).

You may want to view the video in Post #1 at link below to see if it is similar to what you experienced.
 
Last edited:

sdrawkcaB

Defender of Light 📶
Advanced User
Lifetime Premium Member
Joined
Mar 6, 2018
Messages
5,237
Reaction score
10,324
Location
Here
TBH I wish escort live transmitted an auto learn database to the detector based on what live has seen in the past and other users lockout passes... if that was a feature then it’d be the bomb...
They're currently collecting data back from the detectors, including the detectors disposition regarding if it was TrueLocke'd (auto learnt) or not, or if it was manually locked out or not.

It is a possibility with the data they are currently collecting, they absolutely could use this data to push community learned false positives back to locations possibly once they feel ready to do so.
 

InsipidMonkey

Essential Monkey
Security
Advanced User
Premium Member
Joined
Mar 22, 2017
Messages
6,905
Reaction score
15,436
Location
New England
A lot about the inner workings of AutoLean has not been disclosed for obvious reasons
What reason do they have to keep the inner workings secret? They've had an exclusive patent on it for almost 20 years.
 

poolmon

Advanced User
Lifetime Premium Member
Joined
Oct 26, 2013
Messages
762
Reaction score
1,209
A little more on AutoLearn . . .

If a band is turned off in Preferences, AutoLearn is also disabled with respect to that Band according to ER.

If the band is turned on in Preferences, but its alerts are suppressed via the Mode Button (eg Auto No X) AutoLearn is not disabled and continues to learn lockouts in the background.

If a band is segmented in Preferences I have not read definitively about its effect on AutoLearn, but would guess that AutoLearn is turned off for that segment. ER would have to verify this.

So, if you run Auto No X (with the X alerts totally suppressed) AutoLearn continues to store X Lockouts in the background whereas if you turn off the X band in Preferences they are not.
Post automatically merged:

What reason do they have to keep the inner workings secret? They've had an exclusive patent on it for almost 20 years.
I guess its like the "quants" on Wall Street.
They all do quantitative analysis but their individual algorithms that work out the detailed analysis and mechanics of the final result are guarded like Fort Knox.

Valentine had its Arrows, Escort has its AutoLearn and Radenso will have its AI. We will end up buying multiple detectors to support the development of new tech for those companies who keep on innovating.
 
Last edited:

Stoopstroop

PSL +5
Intermediate User
Joined
Jan 20, 2017
Messages
558
Reaction score
502
What reason do they have to keep the inner workings secret? They've had an exclusive patent on it for almost 20 years.
Agree! If escort was not secretive about everything they might gain some trust with their customers. Well that and ACTUALLY fix the problems (lock ups, silent lock up, hard lock ups and laser falsing) with their detectors and just not tell us it is fixed. That would go a long way.....
 
Last edited:

Trinity

Learning to Fly
Beginner User
Joined
Sep 10, 2019
Messages
120
Reaction score
143
I thought it can see the DNA of a signal and know it's not a real police gun just by the way it's signal is being transmitted. My 360 has locked out speed signs on first passes before or sometimes it alerts are strange or very very short range on speed signs.

On real police sources it's spot on though so I don't worry about the way it detects these speed signs.

Escort so called dna tech is nothing like radenso DNA new detector coming out, but escort has claimed they used DNA technology for years now.

Who knows, but I wouldn't worry about it. ¯\_(ツ)_/¯


What you do have to worry about is it locking out real police on accident. Just have to know your normal lock out areas and hopefully you realize something is not right. I passed by locked out police cars before shooting radar and I did have a oh crap moment, but it's rare. You learn from those experiences, but they usually happen close to your already locked out real areas.
 
Last edited:

DrHow

Going “Plaid” ASAP (Tesla S) RDT refugee
General User
Lifetime Premium Member
Joined
May 18, 2018
Messages
3,221
Reaction score
5,361
I agree with a few comments. No engineering data to back up. Only Street use seeing how my devices operate.

I also guess/believe that EL reports back Auto lockouts, frequencies, and possible patterns. That some of those validated are pushed back out.

I do guess/believe it does look at “DNA” of a signal and makes a decision. Including not alerting to RF in a LEO band, initially signaling a alert, sees a behavior of RF (freq Hopping speed sign, or whatever is odd in the RF seen) decides on its own to lock out, or not alert at all for that RF signal.

For example, this speed sign the OP is talking about. Let’s say there is no database or signature, or FPGA like work doing live analysis. It is just reacting to the auto lock out function in basic form. Since the signal is strong, it will pick up the signal farther away than a door opener. And let’s say for the moment that speed sign radar is not 100% fixed frequency when it is transmitting. If varies due to loose tolerances. Or by design (FM or hopping). I could see where the transmitter pops in and out of a certain frequency multiple times while traveling to it. Which might trigger the auto learn three strikes and it gets a lock out. OR, there is more intelligence going on beside a lockout trigger.

That FPGA and DSP is doing something.


Or, as many would suggest, cockamamy firmware that can’t be trusted. I do not subscribe to this theory. The device is up to something, trying to figure out what is real, and what is not.
 
Last edited:

timark251

PSL +5
Intermediate User
Joined
Aug 16, 2017
Messages
502
Reaction score
832
Location
Wayne, IL
I agree with a few comments. No engineering data to back up. Only Street use seeing how they operate.

I also guess/believe that EL reports back Auto lockouts, frequencies, and possible patterns. That some of those validated are pushed back out.

I do guess/believe it does look at “DNA” of a signal and makes a decision. Including not alerting to RF in a LEO band, initially signaling a alert, sees a behavior of RF (freq Hopping speed sign, or whatever is odd in the RF seen) decides on its own to lock out, or not alert at all for that RF signal.

For example, this speed sign the OP is talking about. Let’s say there is no database or signature, or FPGA like work doing live analysis. It is just reacting to the auto lock out function in basic form. Since the signal is strong, it will pick up the signal farther away than a door opener. And let’s say for the moment that speed sign radar is not 100% fixed frequency when it is transmitting. If varies due to loose tolerances. Or by design (FM or hopping). I could see where the transmitter pops in and out of a certain frequency multiple times while traveling to it. Which might trigger the auto learn three strikes and it gets a lock out. OR, there is more intelligence going on beside a lockout trigger.

That FPGA and DSP is doing something.


Or, as many would suggest, cockamamy firmware that can’t be trusted. I do not subscribe to this theory. The device is up to something, trying to figure out what is real, and what is not.
The weird thing is that i can unlock it, then it will continue to alert as you would expect. But within 15-20 seconds it locks it out again. I'll try to get it on video next time I drive by it. Mind you there are multiple signs on this roadway too, about 3/4 of a mile apart going eastbound and westbound. All 6 are between 24.130 and .180 So the detector is actually picking up multiple signs at once, but only locking out the one im approaching. I will say, it's a show watching the arrows go nuts.
 

DrHow

Going “Plaid” ASAP (Tesla S) RDT refugee
General User
Lifetime Premium Member
Joined
May 18, 2018
Messages
3,221
Reaction score
5,361
The weird thing is that i can unlock it, then it will continue to alert as you would expect. But within 15-20 seconds it locks it out again. I'll try to get it on video next time I drive by it. Mind you there are multiple signs on this roadway too, about 3/4 of a mile apart going eastbound and westbound. All 6 are between 24.130 and .180 So the detector is actually picking up multiple signs at once, but only locking out the one im approaching. I will say, it's a show watching the arrows go nuts.
That will be cool to see. A speed sign hell. Picking up more than one at a time, trying to figure out what to do. Looks like it is seeing differences beteeen them and not locking out a huge frequency swath (your example of locking out one, seeing the other not locked yet).

once you get done with the testing, I suggest either locking them out yourself, or letting them be locked out by the device. Makes for a quieter ride. If you want to verify your detector is working, get a used portable LEO radar to test while moving. Ka works best for that so, you do not lock out your radar within a small GPS area while testing. :)
 

Discord Server

Latest threads

Latest posts

Forum statistics

Threads
83,595
Messages
1,270,105
Members
21,102
Latest member
tworightfeet
Top