Long story short, for our Bachelor's thesis, we are proposing a way to map out the roughness/smoothness of a road using proximity sensors, specifically, the Vishay VCNL4200 Proximity Sensor.
The project goes like this:
An array of proximity sensors are mounted on the back of a vehicle like a bike rack;
They are mounted so that they form a line that is perpendicular to the road;
After calibration, and when the vehicle reaches a certain speed, the sensors begin to record deformations in the road, i.e. potholes, cracks, etc;
This data is then sent to a computer to be tabulated in an MS Excel file or some other type of file;
At the click of a button, a graph is then drawn using supplied data graphically showing the state of the road, and;
The computer outputs a percentage of the condition of the road.
This is basically what the project is in a nutshell, it acts like a scanner but instead of scanning what is on the road, it scans its condition and how much it needs repairing. A possible implementation of this is a way to evaluate and prioritize roads that need to be repaired.
Is this type of project feasible? Are there other projects like this that are available? Is there a better component I can do this with? Is there another method of doing this, but more efficient and reliable?
Specifications:
Resolution: 1mm
Distance between each point that is recorded: as small as possible, maybe 1mm also
Road length: 10-500 m
Don't worry about stabilization as that comes later in the planning process. I've seen the Ping))) Ultrasonic Distance Sensor but according to my understanding, it spits out the length of the closest obstacle it can detect.
We could use that here in MA, the roads are crap after every winter! 1mm seems kind of small, maybe 5mm or 10mm instead. We have places with potholes, depressed manhole covers, that are inches deep. Really irritating to drive on with runflat tires, have to swerve constantly to avoid them. Places where they blow out the cracks in the road and fill with rubbery stuff, that stuck up higher than 1mm too. Like a giant rumble strip. And any newly paved road lasts about 2 weeks before they put a trench across it to fix someone water piper or something, and never pave it fully so it turns into a giant rut across the road. Same for internet companies, only it leaves a smaller rut. Don't know why that do hydroscopic (water) boring under the road instead to install new cable.
Sensor placed similar distance apart also, like 10mm.
AWOL:
How are you going to ensure that adjacent (or even close) sensors don't interfere with each other?
Oh, I worded it like an idiot. I mean that for each sensor, the distance between each point that it records is as small as possible, maybe 1 or 5 mm, so, every 1 or 5 mm, it records the distance between it and the road. But for the sensors themselves, depends on the minimum effective distance where they don't interfere with each other.
I don't know much about sensors, so any input of anyone that knows about the correct type or kind of sensor that we need to use is much appreciated.
MorganS:
Reading the datasheet for that sensor it seems more like a brightness detector. It doesn't measure distance directly.
Yeah, I took it more like a starting point since it has the, what I think is, the correct range of distance. But anyone else, feel free to suggest other sensors.
Taking a measurement every 1 mm will be time-consuming. If your sensor is capable of 1000 readings per second (the datasheet says no) then your maximum scanning speed will be 1m/sec. That's slower than walking pace. It will take a long time to scan a reasonable amount of road at that resolution.
Rather than 100 $3 sensors, it might be better to use one $300 sensor. Look up "scanning lidar".
MorganS:
Rather than 100 $3 sensors, it might be better to use one $300 sensor. Look up "scanning lidar".
I looked up LiDAR systems, and it seems as if we need extra resources in order to make it work, like training, extra programming, etc. LiDAR is great, don't get me wrong, it 3D maps the surface or anything the laser sees. But the way I see it using distance sensors are better, in my opinion.
I found this VL53L0X ToF Distance Sensor on polulu.com, it fits my specifications, it has a resolution of 1mm, it is compatible with Arduinos, and it uses the I2C interface. But I think that the typical sampling rate of 60 Hz might be too slow for my liking.
If you're trying to put a lot of these in a row, all looking at about the same area then each one will see part of the reflected light intended for its neighbor. Ultrasonics also have the same problem. That can mean that the 60Hz sample rate is distributed over the entire array - 60 of them can only sample at 1Hz each.
But light moves really fast. If you can ensure that there's more than a nanosecond difference in the timing of each sensor's outgoing pulse then they won't interfere with each other.
We have settled on the VL53L0X, with its 50-60 Hz sampling rate and 1mm resolution. But another challenger has appeared! The LiDAR Lite v3, with a 10mm resolution at 0-2.5m, and 25mm resolution at ranges of 2.5m and above. The thing that strikes me most about it is that it is I2C compatible and it has a sampling rate of >1KHz. Can this be a replacement for the VL53L0X? The thing that I worry about is that (I think) the data recorded by the LiDAR is in 3D, instead of the 2D of the VL53L0X. How will I be able to process it with the Arduino? The International Roughness Index (IRI) measures the change in height of the road on a line and divides it with the distance from the initial point.
Can anyone have any input on this? In terms of measuring road height in a line, what do you think is the better option? The LiDAR Lite v3, or the VL53L0X?
Is that "bike rack" mounted to the axle so it is always the same height (disregarding tire squishing) or on the car body which might bounce up and down several inches or cm?