Go Down

Topic: Hiring arduino expert (not a joke) (Read 1 time) previous topic - next topic

zwieblum

Wen I made a panaorama image device for a friend, we went to using a mechanical wire trigger on Nikon. IR is fine, but not that reliable (and you have to mount it somewhere where the camera sees it)

Riva

3. Want a trigger that is right next to DSLR camera. Not a trigger that is wirelessly tethered to camera that is some distance away. I'd like trigger and camera very close to each other.
The camera would remain at your house just a small (hidden) wireless transmitter near the point you want to detect the car that when triggered sends a signal to a receiver next to your camera and the receiver triggers the camera shutter.
Don't PM me for help as I will ignore it.

pathfinderwy

Yes, electronic trigger looks right. 

Lidar speed guns have a range of 1000m+, so it would seem possible to detect cars at 50-100m, right? I mean, here is a module that detects up to 180m:

https://www.sparkfun.com/products/15180

Nice idea but I can't do an induction loop in the road. 

wvmarle

In ideal circumstances, it can get that range. Certainly not in the rain or snow.
Quality of answers is related to the quality of questions. Good questions will get good answers. Useless answers are a sign of a poor question.

zwieblum

Speedguns (that sort used by pocice) are expensive and the opening angle is bigger than your defined 0.6°. And the operation proncipal is not the same as in your application. You need a kind of switch for object moving perpeniculat to your line of sight, not parallel to your line of sight.

Please look at the datasheet of https://www.sparkfun.com/products/15180: 180m@90% reflectivity,70m@10% reflectivity / 130m@90% reflectivity&100Klux, 50m@10%reflectivity&100Klux --> You will detect new white cars at 100m distance in clear weather.


zoomx

Just other questions
After resolving the car detecting task there is the photo task.
Which lens do you use?
Camera will be fixed? Car arrive from one direction only?
What about a photo taken in night? Long exposure mean a blurry car image.
What about photo taken in bad weather condition? In heavy rain, for example.


wvmarle

What about photo taken in bad weather condition? In heavy rain, for example.
Just mount some wipers on the camera lens, what's the problem?
Quality of answers is related to the quality of questions. Good questions will get good answers. Useless answers are a sign of a poor question.

zoomx

What about heavy snow? Or simply fog?

pathfinderwy

Don't worry too much about details. I wanna pay someone $1200-$1500 for help prototyping this thing. I have reasonable expectations around performance and it does NOT need to be perfect.  If interested and experienced, shoot me a message. thanks!

bobcousins

Don't worry too much about details.
Well, engineers have to worry about details, because if the details are wrong, then it doesn't work => unhappy customer.

For example, a range finder will return the range of whatever is pointed at. It can't tell the difference between a car, a fence, etc. Something like the TF03 may be able to reliably detect the difference of a few meters, maybe not.

Looks like a fun project, I might even get a TF03 to play around with. Either way, this is really a research project.
Please ask questions in the forum so everyone can benefit. PM me for paid work.

wvmarle

The devil is in the details. That's why details matter - and where management and engineering always clash.

The one detail that is still missing from here is how to actually detect those cars from that distance, allowed false positives, allowed false negatives, etc.

All I know is you can't put an induction loop in the road. Well, seriously, that comes as no surprise. But unsure to me is what you CAN place at/near the road, if anything at all.
Quality of answers is related to the quality of questions. Good questions will get good answers. Useless answers are a sign of a poor question.

J-M-L

#26
Jun 16, 2019, 11:52 am Last Edit: Jun 16, 2019, 11:58 am by J-M-L
As the end goal seems to take a picture 100m away, weather and lighting conditions should matter much, right? If you were To correctly detect the presence of a car but can't take a good picture, what's the value of the information?

If that's a right assumption I would build a system based on a video constant scan of the precise location (requires great optics) where you want to see a car and use Model based AI to recognize a car which would then trigger the photo capture from another camera which would have the right aperture, zoom etc...

This probably could run easily on a modern smartphone (or possibly a RPi) connected to a Input video stream as they embed google tensorflow - with good quality and enough training the code could even tell you the brand of the car :)


The LiDAR approach if you 'hit the road' with an angle (ie if the signal does not disappear on the horizon) will not be super  as at that distance i doubt you get the precision to measure the change in hitting the car versus the road (+/- 1m precison at 100m means potential 2m variations between measures)
Hello - Please do not PM me for help,  others will benefit as well if you post your question publicly on the forums.
Bonjour Pas de messages privés SVP, postez dans le forum directement pour que ça profite à tous

zwieblum

Na, that would be too easy. Just imagine, juo could simply do a "apt-get install motion" and had a ready to go soluton. No way!

zoomx

Taking a good picture 100m away during night can be very expensive.

Go Up