1

For this project we need to count weels of a car (if possible at all), one side only. The scenario is within the following scope:

  • The car is always passing in a straight line
  • It is cars only, so all other traffic can be ignored
  • Road type is asphalt. Bricks, macadam or concrete detection is optional
  • Distance to the car will be around 4 meter (13 foot)
  • The tires will have a deviation to the measurement device of 20cm (8inch)
  • The scope is set to car tires of maximum 20" in diameter
  • At maximum speed of 130km/h or 80mph
  • Minimum speed can be any, but cars will never stand still, they will eventualy move (traffic jam)

The method of measuring should comply with the following rules:

  • No visible light sender or receiver can be used
  • Interaction with the tire is not possible (example: a cable to run over)
  • The measurement device can become dirty
  • Must be all weather type
  • Power consumption should be as low as possible
  • Device can be in sleep mode
  • Size should not exceed 50x50x50cm (20x20x20inch)

Since we cannot come up with possible measurement methods i'm looking for inspiration on how this might be achieved. Even if it might not work but it is viable we will look into it.

Michael Beuving
  • 11
  • 3
  • Are the remarks in *italics* actually part of the requirements? Because "it can get dirty so it doesn't work anymore" applies to a lot more than visible light, and visible light can be way robust against different light conditions, for example. – Marcus Müller Jan 30 '19 at 10:03
  • At any vehicle speed? – Dorian Jan 30 '19 at 10:18
  • Yes it is a requirement, i just gave the example in italics. The challenge for us is to build something without visible light. Since pointing a camera would be the easiest way to achieve, it is not what we want. – Michael Beuving Jan 30 '19 at 10:18
  • @Dorian i've updated the question, its max 130kmh – Michael Beuving Jan 30 '19 at 10:20
  • 1
    And minimum speed? – Dorian Jan 30 '19 at 10:21
  • Can be any, car will always pass (since they cannot park on the road), the duration is just not always the same (traffic jam for example), or slow driving behind a truck. – Michael Beuving Jan 30 '19 at 10:23
  • You said max 20", but is that the diameter, or merely the width of the tread? What would be the minimum tread width? Are we detecting motorcycles? Are we trying to filter out trailers (meaning something a mowing service would pull)? Do we have to do this through macadam (tar-and-crushed-rock)? – MicroservicesOnDDD Jan 30 '19 at 11:20
  • Any vehicle ? Motorcycle, trucks with more than two wheels on the same axis? – Dorian Jan 30 '19 at 11:23
  • The question specifically says "car". Why do you think the answer would be anything other than two? Also, 20" is awfully tiny for a car tire -- what is supposed to happen when larger tires go by? Or are you referring to wheel diameter? – Dave Tweed Jan 30 '19 at 12:23
  • It is cars only, and 20" is indeed the diameter – Michael Beuving Jan 30 '19 at 12:24
  • @MicroservicesOnDDD i've updated the question – Michael Beuving Jan 30 '19 at 13:04

2 Answers2

1

Three ideas spring to mind:

  • Doppler radar should work well; one can pretty reliably detect the microdoppler caused by the rotation of wheels with a directional antenna (and preferably a high frequency microwave radar)
  • acoustics should be pretty reliable, too, given a bit of signal processing
  • vibrational analysis (which is essentially, acoustics) on the concrete should work just as well.

Other than that, a simple photoelectric through-beam sensor would work; with a rain / dirt shield, that should be pretty reliable too. I don't fully buy "light can't work in dirt", as that's a kind of sensing you find in industrial environments extremely often.

Marcus Müller
  • 88,280
  • 5
  • 131
  • 237
  • Doppler radar sounds interesting, that is something we will be looking into. Do you know of a working device/project that does something similar with Doppler radar? – Michael Beuving Jan 30 '19 at 10:22
  • Well, there's beyond-the-hill helicopter detection based on Microdoppler, if you're looking for detection of actually rotating things. Doppler Radar in general is used in every speed trap, automatic doors, flight radars, particulate flow sensing... – Marcus Müller Jan 30 '19 at 10:40
0

Standard method used in Canada for controlling traffic lights or vehicle-counting is to use inductive magnetic sensor loops buried in the asphalt road surface. These loops are fairly large - a rectangle or square several feet in each dimension.

There are two distinct uses of vehicle detector loops that I see: traffic light control and vehicle counting on freeways or other high-volume locations.

Those loops used for traffic control appear to be either a single loop that encompasses all active lanes of traffic or one loop centered on each active traffic lane.

Most of the freeway installations that I have seen have two loops in each traffic lane where the loops are separated from each other by several feet. Although I do not know the exact reason they do this, I believe it allows the system to work accurately even if there is a traffic jam or other type of slow-down.

Dwayne Reid
  • 23,390
  • 2
  • 35
  • 67
  • The requirement is to count wheels, not vehicles. Specifically, wheels on one side of cars. Since cars with anything other than two axles are extremely rare, perhaps it is sufficient to say that wheels = vehicles * 2. – Dave Tweed Jan 30 '19 at 12:13
  • 1
    Cars with more than two axles are not _extremely_ rare. Trucks will quite often be three axled, and semis might have three axles just on the tractor unit. – MrGerber Jan 30 '19 at 13:50