Introducing Radenso Theia powered by Rai (Video Links and FAQ on first page)

Riptide

Recovering M3 Addict
Advanced User
Premium Member
Joined
Oct 29, 2010
Messages
4,192
Reaction score
2,138
I believe I recall hearing in one of the video's where Jon said they have 80 2080s running? Or maybe 80 rack spaces. And a 2080 is 75% of a V100 iirc. Crazy horsepower either way.
Well damn, nevermind then lol
 

tempnexus

Breaking The Social Norm One Post at a time
Banned User
General User
Premium Member
Joined
Jul 31, 2015
Messages
6,276
Reaction score
7,796
Age
108
Location
DMV

CobawLT2010

Doesn’t time fly when you’re waiting two weeks?!
Advanced User
Premium Member
Joined
Oct 26, 2011
Messages
7,268
Reaction score
3,747
Location
Sick of here...
From what I understand, less expensive models with trimmed features will be developed after the initial release.
Just been reading here the last few days, liking posts etc. First post in this melee...
Thanks @Jon at Radenso for providing the kick in the butt this industry needs!! Looking forward to the tests and how the unit turns out. I'm certainly enjoying explaining it to my GF (who runs an R3) - Good things ahead
I’m not even sure I NEED this tbh... my R3 fits my needs just fine.

I get almost no falses on my commute already. Maybe one a week tops...
 

synistyr

I need more coffee
Advanced User
Lifetime Premium Member
Joined
Aug 4, 2014
Messages
1,899
Reaction score
2,074
If Rai isn't busy right now, could you get her to write some Pro M filters for those mother effin' Mazda BSMs?? :)
 

mz2slow

Learning to Drive
General User
Joined
May 23, 2019
Messages
10
Reaction score
25
I'm using the standard definition for the word "Signal", the same one used in any discussion about RF Energy in the microwave bands. Basically RF Signals are emissions from transmitters that have such properties as wavelength, frequency, amplitude, and phase. In identifying a "Signal" we use all of these "Identifiers" to allow us to put a "Origins" label in said Signal. In additional to these minimum basic properties of a Signal they can have any number of other "Identifiers" based on the purpose of the signal, its transmitter operational type, the mode of Modulation, its carrier and/or subcarriers, in addition the signal exhibits signature characteristics of the components that make up the transmitter such as the LO for example.

And yes in that definition of Signal is the word "Frequency". Theia could use Frequency in Identifying a Police Radar Signal, but all I have been attempting to get across is that it could just as easily be designed not to need or use a Signals actual Frequency in that Identification (and that from my standpoint, I hope it doesn't even though in not using it, it could still display it). So if one is truly actually Identifying Police Radar Signals, I am of the option that Frequency should be treated as a constant because the device has already been designed only to look for Police Radar within those "Frequencies" where Police Radar are. In my opinion this has multiple pluses and not one single drawback. First it makes Identifying quicker and easier by eliminating a variable. Next Theia could analyze ALL Signals the device allows past its frontend and Receiver to Identify them for what they are. This guarantees capture and Alert of any and all actual Police Radar, no matter how far off center said Radar Gun is (what people call out of tune guns). And as a fail safe Theia would be analyzing all signals within the range completely independent of a Signals actual Frequency.

In using actual Signal Identification as the requirement for Alerting only to Police Radar and of course Alerting to any signal that can't be classified or identified as a Fail Safe IE better to get an Alert and brake then get a Ticket.

So for those people that were thinking I was saying Frequency isn't part of this at all, I was not. I am saying absolute Frequency accuracy isn't a requirement nor does Frequency even have to be used if the device is designed specifically to only Identify Police Radar inside and whatever safety margin outside of their three allowed FCC US frequency bands. So what I am saying is I would rather have frequency be a constant in my matrix of Identifying any and all Police Radar Signals though bandpass and receiver design so that ALL signals that get through my Receiver are analyzed and Identified as Police Radar causing an Alert or Identified as not Radar without an Alert, independent of Frequency! Because again Frequency is already defined by range within the design of the device.

Take an FM Radio for example, do I really care what the Frequency Range is, no, I just want to be able to listen to music. Do I really care what the Frequency of the Radio Station is, no, except if I know it and want to input that Frequency directly instead of searching for it. Now you may have just said to yourself "Exactly", but that is NOT a gotcha because unlike the FM Radio where I want a specific station and I am not searching, I WANT my Radar Detector always searching not just looking for Police Radar at exactly 35.500 GHz! So again if Frequency Range is defined by the Device's design (FM Only Radio will not receive AM), then in the design of the true Radar Detector it has also been defined which can then make using Frequency to Identify a Signal optional.

Sure "Knowing" that Frequency allows said Police Radar Signal to be classified further as in let's take the Alert of "Radar Falcon HR Front" which can further be identified as K band and further identified as 24.126 GHz. Do I personally need all that, no longer not with Theia. Once I hear Radar Front, I am on the brakes, the rest of it is just nice additional information but doesn't change how I react to it. Now with Radar unlike Lidar which when we hear an Alert it is auto drop the Anchor time and knowing it is Dragon Eye or TrueSpeed S after the fact doesn't change our reaction to the Alert, with Radar and the Theia, I would love to have the option to customize the Alert tone based on Identified Device versus Band! Again just like ALP with Lidar Identifying the device, with Theia Identifying the Device, I would personally like my most urgent tone to be on All I/O and the K Band low power Falcon HR as my response to both needs to be the same IE with the Theia I no longer have real use for separate K Band and Ka Band Identifying Tones. I would rather my Tones define the Urgency of my reaction to specific Alerts no matter the Band. Again moving away from the need to even bother knowing Frequency. Does this help in understanding why and how Frequency can become a constant instead and why it might no longer be needed for Alerting, Identifying, or even Tones?
.
.
Thank you. I agree that in the UI, a well performing system as envisioned by Radenso here does not need to relay frequency information to the user.

I agree with most of what you wrote. My only point was that writing "frequency is irrelevant", as many have done, is technically inaccurate, and very misleading to those who understand in less depth. The phrase "signal without frequency" in the domain of RF discussions is completely illogical. Yes, one can back out the carrier frequency and look only at bandwidth (max - min frequency), or frequency change over time. However, measuring frequency is critical to these things too! We should be careful to say things like "carrier frequency doesn't matter", rather than "frequency doesn't matter". That's all I was trying to say.

That, and that my questions about accuracy and neural net training are highly relevant, since the signals Rai identifies are affected by environmental factors and noise / interference that affect precisely the frequency components of any signal.
 

RedRocket

Lidar Shootist
Advanced User
Premium Member
Joined
Apr 10, 2012
Messages
4,264
Reaction score
4,870
Location
USA
Should be just under USB PD requirements (tried to avoid that at all costs to eliminate complexity).

Yes, we use Tensorflow for our AI. We train it on a bunch of RTX 2080 TI video cards.
How many bits your ADC runnin' ?
Jon - I got a notice you quoted my Post above, but I do not see your answer anywhere(ans. was supposed to be in your Post #455)...did you delete it for some reason ?
 
Last edited:

CobawLT2010

Doesn’t time fly when you’re waiting two weeks?!
Advanced User
Premium Member
Joined
Oct 26, 2011
Messages
7,268
Reaction score
3,747
Location
Sick of here...
Thank you. I agree that in the UI, a well performing system as envisioned by Radenso here does not need to relay frequency information to the user.

I agree with most of what you wrote. My only point was that writing "frequency is irrelevant", as many have done, is technically inaccurate, and very misleading to those who understand in less depth. The phrase "signal without frequency" in the domain of RF discussions is completely illogical. Yes, one can back out the carrier frequency and look only at bandwidth (max - min frequency), or frequency change over time. However, measuring frequency is critical to these things too! We should be careful to say things like "carrier frequency doesn't matter", rather than "frequency doesn't matter". That's all I was trying to say.

That, and that my questions about accuracy and neural net training are highly relevant, since the signals Rai identifies are affected by environmental factors and noise / interference that affect precisely the frequency components of any signal.
Dude... let’s let these shills be shillin and let these kool aid drinkers get drunk on it...

Trust but verify... there is always more than one way to achieve a desired end goal.

for me it’s all down to tests and price... I will wait until the bugs are worked out... everything works and verified... and I can justify the cost.

I never buy into hype... my expectations are always low... because then I am NEVER disappointed. A LOT of my questions have not been answered...
 

mb300sd

PSL +200
Intermediate User
Premium Member
Joined
Mar 14, 2012
Messages
507
Reaction score
434
Location
GA
Should be just under USB PD requirements (tried to avoid that at all costs to eliminate complexity).

Yes, we use Tensorflow for our AI. We train it on a bunch of RTX 2080 TI video cards.
Good to hear, if it's not PD, then it should be pretty easy to find a tiny PCB that can be extracted from something and fit up inside the headliner. You have no idea how long it took me to find a tiny QC3.0 adapter that I could gut in order to hardwire into my car and not use the built in slow charging ports or taking up the 12v. Of coarse they're all over Amazon 2 years later so there's probably nothing to worry about.
 

surprisinguy

Government can only give what it takes...
Advanced User
Premium Member
Joined
Mar 10, 2015
Messages
1,331
Reaction score
1,291
Location
North Carolina
I’m not even sure I NEED this tbh...
I definitely "need" some type of radar CM. Being able to afford the new Radenso might not even be an option for me. IF a new V1 is introduced and is in its current price range +100 or so, it might be just fine for me if its testing pans out. Heck it might even debut before the new Radenso (2 weeks, lol), but who knows at this point. I'm used to the JBV1 combination and @johnboy00 has said he'll support the new model if and when it comes out.
I'm sure the new Radenso will be incredible though - looking forward to testing when available.
 

mb300sd

PSL +200
Intermediate User
Premium Member
Joined
Mar 14, 2012
Messages
507
Reaction score
434
Location
GA
Jon - I got a notice you quoted my Post above, but I do not see your answer anywhere(ans. was supposed to be in your Post #455)...did you delete it for some reason ?
Also interested in ADC specs. If they're not too high spec, we could get community members with SDRs to record new guns and BSMs for training.
 

thefrog1394

Cruise Control PSL+9
Beginner User
Joined
Nov 8, 2018
Messages
241
Reaction score
233
Location
Ohio/PA
The issue with USB power is needing a separate adapter to hardwire, sometimes a massive pain when 12v is available in the headliner. I hope it has RJ11 or supports 12V on the USB port.
@Jon at Radenso,

Can you give us an idea of the power requirements at this time or is it to early in the development cycle to know for sure?


Looks like I'll need to start measuring for a USB C cord to run the headliner, down the pillar and tapping into the ALP/dash cam source...
You can also just grab a small 12V->5V adaptor and tuck it up in the headliner and hardwire to either an existing 12V source or run 12V down the pillar which would likely be cheaper and easier.
 

InsipidMonkey

Premium Monkey
Advanced User
Premium Member
Joined
Mar 22, 2017
Messages
6,077
Reaction score
12,702
Location
New England
Also interested in ADC specs. If they're not too high spec, we could get community members with SDRs to record new guns and BSMs for training.
In order to record new guns and BSMs in a useful way to train the AI, you'd have to use the same RF front end that will be feeding the AI in the detector. Even if you could downmix the K/Ka band signals to something a consumer-grade SDR could receive, the actual data fed to the AI would be different and not particularly useful for training.

Radenso's new detector contains two completely new approaches that we haven't seen before in a RD. I'd argue the more important part is the new single-conversion RF receiver and I/Q demodulator, as this enables much higher quality data (a "higher resolution picture" if you will) to be fed to the AI for inferencing. The AI itself is exciting and novel, and it's possible that other detector manufacturers that have FPGAs or powerful processors in their current or future detectors could leverage similar AI inferencing. The performance in those cases will always be limited by the RF design, however, as current generation superheterodyne receivers provide much less data (eg they can't see phase) which gives the AI less information to make inferences from.
 

mb300sd

PSL +200
Intermediate User
Premium Member
Joined
Mar 14, 2012
Messages
507
Reaction score
434
Location
GA
In order to record new guns and BSMs in a useful way to train the AI, you'd have to use the same RF front end that will be feeding the AI in the detector. Even if you could downmix the K/Ka band signals to something a consumer-grade SDR could receive, the actual data fed to the AI would be different and not particularly useful for training.

Radenso's new detector contains two completely new approaches that we haven't seen before in a RD. I'd argue the more important part is the new single-conversion RF receiver and I/Q demodulator, as this enables much higher quality data (a "higher resolution picture" if you will) to be fed to the AI for inferencing. The AI itself is exciting and novel, and it's possible that other detector manufacturers that have FPGAs or powerful processors in their current or future detectors could leverage similar AI inferencing. The performance in those cases will always be limited by the RF design, however, as current generation superheterodyne receivers provide much less data (eg they can't see phase) which gives the AI less information to make inferences from.
I'm definitely more of a beginner in the SDR space, but it's fun to play with. Didn't think of the front end issues, but also the fact that the detector itself could be used as the recording device if Radenso wanted community recordings.
 

Zhe Wiz

Learning to Drive
General User
Joined
Jun 30, 2019
Messages
4
Reaction score
5
How hard would it be for radar gun MFRs to create a variable signature gun that can't be "learned" by Rai? I don't know enough about the technology to answer that question. Is it even possible? I worry the release of Theia might create a change in radar guns that can't be accounted for using Theia/Rai's methods. Also, why don't police simply move to Laser, rendering radar detectors (even Theia) obsolete? I know once upon a time laser was prohibitively expensive and so many departments couldn't afford it. Is that still true? As many have pointed out, laser detectors are essentially useless and many of us don't really want to go with jammers due to cost and (relatively) complex installation issues.

Curious and playing devil's advocate.
 

DChiJEllis

Lifetime RDF Contributor
Moderator
Advanced User
Lifetime Premium Member
Joined
Jul 31, 2016
Messages
1,875
Reaction score
2,922
Location
Trollstigen, Lyseveien og Atlanterhavsveien
@Jon at Radenso @Randy at Radenso If you want to run this against some Gatso and do some signal recording, let me know. Be happy to show you around this area and where/how they are being used.
 

Riptide

Recovering M3 Addict
Advanced User
Premium Member
Joined
Oct 29, 2010
Messages
4,192
Reaction score
2,138
How hard would it be for radar gun MFRs to create a variable signature gun that can't be "learned" by Rai? I don't know enough about the technology to answer that question. Is it even possible? I worry the release of Theia might create a change in radar guns that can't be accounted for using Theia/Rai's methods.
Not enough people with RDs in their car for them to care honestly, far fewer will have the Theia.

Also, why don't police simply move to Laser, rendering radar detectors (even Theia) obsolete? I know once upon a time laser was prohibitively expensive and so many departments couldn't afford it. Is that still true?
I would love if every radar gun was thrown out and we could all slap ALPs on our cars and call it a day, but that won't happen.
Two reasons off the top of my head: Lidar can't currently be used from a moving vehicle, and lidar requires much more attention/work from the police officer (can't just turn it on and then sit back watching the speeds).
 

DChiJEllis

Lifetime RDF Contributor
Moderator
Advanced User
Lifetime Premium Member
Joined
Jul 31, 2016
Messages
1,875
Reaction score
2,922
Location
Trollstigen, Lyseveien og Atlanterhavsveien
How hard would it be for radar gun MFRs to create a variable signature gun that can't be "learned" by Rai? I don't know enough about the technology to answer that question. Is it even possible? I worry the release of Theia might create a change in radar guns that can't be accounted for using Theia/Rai's methods.
Watch @Vortex's new video
 

RedRocket

Lidar Shootist
Advanced User
Premium Member
Joined
Apr 10, 2012
Messages
4,264
Reaction score
4,870
Location
USA
Also interested in ADC specs. If they're not too high spec, we could get community members with SDRs to record new guns and BSMs for training.
My interest wasn't your's for inquiring about the capability of the ADC, it was to ball park the dynamic range number. That number is critical in a garbage polluted microwave band where you've got nearby much stronger BSM/ACC systems in adjacent lanes on the highway that could be causing fundamental overload & de-sensing the receiver Front End causing the much weaker Police radar ahead to be completely missed !
As @Jon at Radenso @Hügel66 are avoiding a reply- here's some guesstimate #'s:
12-bit...~100db
16-bit...~Obscene (lol)

Sure would like to hear some SNR numbers, too.
 

Vortex

Making Videos
Observer
Advanced User
Lifetime Premium Member
Joined
Jul 19, 2012
Messages
18,935
Reaction score
43,179
Location
Washington State

Discord Server

Latest threads

Forum statistics

Threads
79,372
Messages
1,208,281
Members
20,128
Latest member
kawkaw2435
Top