I'm hoping the expertise and experience on this forum can help me out on a project I'd like to undertake. Please keep in mind that I'm neither a developer nor electronics engineer. But am somewhat technical and I can follow instructions.
I'm looking to have a device that will log vehicles passing by. In the end, all I really need is an output with the time stamp of each passage that I can place in Excel for analysis. I've seen all sorts of different types of sensors available from optical, microwave (Doppler radar), sonar, loop/inductance, and the like. I really have no idea what might be the best for my situation but the options and sensors available seem at first glance to lend themselves very well to an Arduino project. (I'm also open to shifting to Raspberry Pi, if indicated.) Here is some high-level information:
I'll need some method of getting the vehicle passage time stamps in Excel.
I'll need to record for at least 36 hours, although 72 (or longer) would be preferable.
I'll need fairly low power consumption, as I'll be running on a 12 VDC Deep Cycle Battery.
Vehicles will most often be traveling between 20 and 90 MPH.
I will be very limited to placing anything in the road (such as a reflector).
I cannot place anything on opposite sides of the road.
I can place equipment and sensors in the median.
I will have the ability to raise the sensor, so it's not just a side shot across the road. I can point the sensor at the lane.
I do not require 100% accuracy, although it needs to be approaching or preferably better than 95%.
On occasion I will have the ability to place sensors overhead (such as on a gantry or overhead sign). But this would only be if I need to gather information on inside lanes.
Most often the lane I'm checking will be the inside (median) lane. But something that would allow me to check inside lanes would be a big bonus.
As I mentioned, I'm not at all sure what route or sensor type would be the best way to go. I'm thinking an inductance loop (such as with gate openers) might be easily achievable. But I was also thinking that a DIY radar gun that outputs the Doppler sound might be better, as it might be very 'point-able'. Then again sonar could be very accurate, but everything I've seen for off-the-shelf is either for close proximity or outrageously expensive.
Right now I'm exploring options as to what might be do-able with a reasonable cost and effort. Any help will be appreciated.
The complicated part of this will be detecting the vehicles. The rest of the project is straightforward.
If I had access to a police car, this would be easy if I could modify their radar system. The question is, can you buy police radar systems? I don't see how one could get into mischief with a police radar, but because it's designed to be sold to the government who isn't know for trying to find "good deals" I would anticipate the price to be somewhat high for a hobby experiment. If you are trying to start a business with this, then I recommend you look into that.
I'd be surprised if their system they bring out that shows you how fast you are going doesn't already record traffic. Have you looked into renting one of those units?
So you can't put a simple reflector on the opposite side but you can build a gantry over the road?
How many lanes? How do you define "95%"? If the inside lane is important and the logger is on the median, do you care if it misses a few vehicles traveling side by side in the outer lane(s)?
The old amphometer rubber tube sensors are still used and still the most effective. I don't know how you get permission to staple a rubber tube onto the road but that would be my choice if you need accuracy on a multi lane road.
Thanks for the responses. Here are a few more details that will hopefully clear up some of the questions.
I'm independently helping some roadways validate their sophisticated systems. Right now I'm doing it manually. I take video of the lane. I play back the video on my PC and every time a vehicle passes I press a key that logs a time stamp (corresponding to the video time). I then compare my log with the system log and identify any discrepancies. The video serves as absolute proof that their sophisticated system missed something. If I could use a sensor to create a log with reasonable accuracy then I would only need to look at the video where comparing their log to mine shows a discrepancy.
The majority of the time I won't be able to put anything on the road (such as a reflector).
For most locations, if it is already there, I can put a sensor on an existing gantry or sign (I won't be building anything).
Almost all of the time there will be multiple lanes of traffic, although most of the time I'll only be examining the inside lane.
KEY: When I log the vehicles it needs to be specific to a lane. (So although I could put a sensor on one side and a reflector on the other side, it won't do me any good. The same is true for a rubber tube.)
When I say 95% accurate, I'm meaning that for every 100 vehicles passing in a given lane I correctly detect at least 95 of them &/or do not 'create' vehicles that were not in the lane such as counting a vehicle in an adjacent lane. (In this case, when I say 95%, I'm not trying to be statistically accurate; i.e. accuracy and confidence using binomial or beta distribution calculations. I'm just saying that if I can't correctly detect 95 or better vehicles out of every 100 passing then the exercise likely isn't worth doing.)
The[ jaycar.us](http://"http://www.jaycar.us/productView.asp?ID=KC5441" &w="Radar&form=KEYWORD") link to the coffee can radar showed out of stock, but I had earlier found one on Amazon (likely the same unit). I'd also found an induction loop on Amazon that looked promising. The radar seemed the most promising as it could be pointed at a specific lane. The induction loop would be limited to just one of the edge lanes.
Let's say I went with the radar kit. It outputs sound. Would I use the sound level to detect the vehicle or would I tap into some other circuit on the board? Once that is decided how would I go about using an Arduino to create a log file? As I mentioned, I have no idea how to go about doing this.
dave-in-nj, I just saw your post. Yes, using existing paint and pointing at the far line marker would work for lanes not in the middle. For the inside lanes, I wouldn't be allowed to glue a reflector in the middle of the lane, but I might be allowed to place a small circle of paint (maybe a 4" circle or less). I didn't know an IR beam would work on paint. That would be an option.
On occasion I will have the ability to place sensors overhead (such as on a gantry or overhead sign). But this would only be if I need to gather information on inside lanes.
Most often the lane I'm checking will be the inside (median) lane. But something that would allow me to check inside lanes would be a big bonus.
what is the difference between an inside lane and an inside lane ?
you will be checking the inside lanes most often
but it would be a bonus if you could monitor the inside lane.
are these local variables and not global ones ?
OK, so you're working for a tollroad type organization and you have access to their gantries over the road. I see many tollroads do have a few squares of paint near the gantry which are obviously reflectors. You really can't paint a small square in the middle of each monitored lane?
Even without the reflector, you could aim a light source at the middle of the lane from above and get something back, even if it's zero. Any vehicle passing through your light beam will reflect more or less, so you really only need to detect changes. An infra-red beam is very easy to make and detect.
That's kind of what the radar detector is doing. It could be pointed at a reflective sheet of metal or up into the sky and it's still looking for changes caused by moving vehicles.
Does passive IR work? A basic burglar alarm type sensor with some shielding around it so that it only 'sees' one lane? Passive IR relies on temperature differences so it may not work well at night when the skin of the car is the same temperature as the air.
Hi,
Can you post/attach a diagram of the road layout, placement of the logging unit and the terminology that you are using for inner middle etc.
For example an over head view.
This will make sure we are all on the same page.
Thanks..Tom.....
PS Use REPLY rather than QUICK REPLY to find the attachment option.
Attached is a sample of a typical roadway that I will monitor.
In most cases I will be limited to placing a sensor or sensors where the lightning bolts are shown. That is, attached to one of the uprights of the gantry.
In most cases I will not have access to place anything on the gantry that is going over the roadway. There may be exceptions, but I'd rather not devise a system that relies upon being directly over the lane.
I will have the option of placing the sensor high on the gantry uprights. In this way, if I have to monitor an inside lane then I'll be able to aim down as well as across. This will mean that a large vehicle (ex: bus) in the median lane could obscure a vehicle in an inside lane. I can live with that happening on occasion.
I've only shown pointing sensors towards oncoming traffic, but I can just as well point at the backsides of vehicles, or I can do both.
In most cases, I won't be able to paint the roadway. However, I was just thinking that I could temporarily stick down a piece of Prismatic Tape in the lane. I would prefer not to count on it, but if it's the best option then I can explore it further.
odometer:
If you try to use your system to validate theirs, you will end up with two systems to validate (yours and theirs).
Yes and that's a good point, but it's not limiting. I'm performing the validation already. The difference is that with what I'm doing now I need to manually review every vehicle. With such a tool, I only need to manually review a statistically relevant number of vehicles (with a passing accuracy for a specified confidence level) and then for all others I will only need to review where there is a discrepancy between the tool and the roadway report. From a numbers perspective, I can validate the tool in as little as 1.2k reviews. Thereafter I just look at discrepancies. Without the tool I will manually review at least ~4k passages and as many as 10k.
While I'm not a developer, I am a hobbyist. Over the decades I've built numerous tools and programs from games and inventory applications in Hypercard to modern tools in HTML5 and JavaScript, PL-SQL, and Excel. I've also designed and assembled purpose built electronic devices using COTS products. This would be my first venture into merging assembly and coding, but I'm up for the challenge. This is something I'd like to do myself; not pay for someone to construct. Also, my employer isn't paying me to do this. I'll make the same salary and my employer is still much more efficient than competitors. I'm doing this on my own personal time as a personal challenge to improve a process when I see the possibility exists - and for fun.
Yet thanks for pointing me to the Mechanical Turk. That is a resource I didn't know existed. I can see it coming in handy from time to time.
If going with active IR (beam returning from a reflective surface), can someone point me to a unit that would be sufficiently fast and is compatible with logging on an Arduino?
