Senior Design project: calculating distance

[jmilone]

Hey guys,
I'm currently a senior computer engineering major at Georgia Tech in Atlanta. For my senior design project we are in need of a method of tracking distance up to around 50 meters. Right now, we are doing X coordinate (left to right) as well as depth tracking using blob detection and a somewhat powerful webcam.

However, we would like to use something a bit more accurate/powerful. We have thought of using RFIDs, ZigBee, laser range finders, active radar, ultrasonic and FM/AM transmitter/receivers. So far we haven't found a cost effective/accurate/easy solution to try ( everything must be < $400).

Our advisor suggested that we try to use radio equipment. Our basic idea would be to setup 1 transmitter on the moving object, as well as 2 stationary receivers at a known distance apart. If we could find out the distance between the transmitter and each receiver, simple triangulation could be used to calculate distance from a reference point.

So my question is this: is there a relatively simple way to calculate distance using radio waves (or other methods)? I've seen a couple threads on here but came to no conclusions. I was thinking instead of measuring time which could be hardware intensive, distance could be calculating by using signal strength (heat dissipation or diode?).

Anyways, if anyone has any suggestions for calculating distance, I would be very appreciative!

Thanks again

 

[vne147]

If we could find out the distance between the transmitter and each receiver, simple triangulation could be used to calculate distance from a reference point.

I don't know if I understand this statement. You find distance by triangulation, you don't need it for the calculation. Distance and direction are part of the solution. Your reference points are the locations of the receivers. Each one will sense a slightly different direction to the transmitter. The farther apart the receivers are placed, the greater the difference in direction. If you draw an imaginary line from each receiver in the direction to the transmitter, where the two lines meet is where the transmitter is.

This method of triangulation and navigation has been used in aviation for a LONG time and is so old in fact that it's now being phased out as obsolete.

Check this out.

 

[bobledoux]

Here are some ideas:

Aviation VOR uses an omnidirectional radio signal followed by a very directional signal that sweeps 30 times a second. The time delay between omni-directional and directional pulses gives relative bearing. Light could be used for the same purpose.

The Seattle Robotics Club newsletter had an article that discussed using reflectors to permit triangulation by a robot. Look in their archives.

Consider a radio or light signal followed by an audio or ultrasonic signal. Delay between reception of the two is a function of the speed of sound versus speed of light.

For a light signal, two side-by-side light sensors, each mounted in the end of a tube, will report the same reading when both are pointed at a distant light source.

Review your trig to see what it takes to solve for the angles and distances in a triangle.

 

[crutschow]

An interesting RF method is to FM sweep the frequency of an RF transmitted signal with a triangular wave (using a VCO). Then you measure the beat (difference) frequency between the transmitted and received wave. The beat frequency is proportional to the RF frequency, FM deviation, and sweep frequency, and the time it takes at the speed of light for the RF to travel from the transmitter and be reflected back. The beat frequency is relatively constant (assuming a linear triangular FM modulation) for a given distance.

By proper selection of the RF frequency, sweep frequency, and FM deviation (a task left to the reader), you can achieve a beat frequency that is readily measured for your distances of interest.

Edit: As an example, if you modulate the RF at an FM rate of 1Hz/ns, then for a target 50' away the round trip time is about 100ns, and the beat frequency would be 100Hz.

 

[tylernt]

I would think some kind of active radar would be less than $400. The remote (mobile) station just needs a simple CW radio repeater -- plug a receiver into a transmitter (use a different frequency for each ) and you're done. Simple ham radio CW circuits for this could probably be built on a small perfboard for the price of a pizza.

The base station only needs a CW transmitter, a stable and accurate oscillator, and something like a PIC (maybe even a Basic Stamp) to trigger the transmitter and then count the number of oscillations that occur until it receives the echo from the remote station. A little calibration will eliminate the error resulting from the small delay inherent in the remote repeater circuit. If your PIC can't keep up with the oscillator, there are binary counter ICs for a couple bucks that could be added. The programmer for the PIC is the expensive part here, but still well within budget I would think.

Don't forget to observe FCC Part 15 rules if you use any kind of RF-based system.

 

[Mr RB]

For 50 metres and <$400 I would use ultrasonics, it should be easy enough and give pretty good accuracies over that sort of distance.

I would have the base station "ping" and the remote device respond.

 

[Roff]

GPS is sometimes used to determine relative positioning.
https://www.electro-tech-online.com/custompdfs/2010/02/ch6.pdf

 

[jmilone]

Thanks for the very helpful advice everyone! Our group spoke to our professor today and he basically summed up some stuff with us:

- Using RSSI to determine location will be very inaccurate as the signal will bounce off many objects (outdoor setting)
- If we come up with some RF solution, he said we should basically drop out of school and start a company doing that because it will make us rich - meaning it will be very hard to do

We also spoke with Linx technologies, who said there would be little support for this kind of project

We came up with some other options:

- Use a webcam to scan the horizontal plane we are covering and locate the object, then aim & send an ultrasonic pulse to the object and receive the time of flight information.
- Set up multiple (3+) webcams to scan the field and have them speak to each other to find the distance to the target

Our budget is <= $400, if there are any sure proof methods we would appreciate them (proposal due Monday :/ )

 

[tylernt]

Our advisor suggested that we try to use radio equipment.

Our group spoke to our professor today and he basically summed up some stuff with us:
- If we come up with some RF solution, he said we should basically drop out of school and start a company doing that because it will make us rich - meaning it will be very hard to do

I think your adviser was right and your prof was leading you down the wrong track. If you can do time domain measurement with ultrasonics, you can do it with RF (and RF doesn't even need the webcams for aiming because it's omnidirectional). I agree RSSI is no good but radar is a well-known tech by now with lots of companies competing in the various industries (military/police/aviation/etc). I.E., RF distance measurement is not magical and there's no big secret that will pave your way to fame if you discover it.

You don't need to recreate a guided missile, you just need a precision time measuring tool connected to some RF transceivers.

 

[Mr RB]

What about putting an IR transmitter on the mobile, then get 2 triangulated sensors with IR sensor mounted on a cheap RC servo? As long as they are far enough apart then can each report an angle to the mobile unit, and you just triangulate distance and position based on that.

 

[superfrog]

laser rangefinder is great but over your budget I believe. If you can manage to get one that fits your budget, you won t regret it.

If you wish to do positionning with two references, I would strongly suggest doing error propagation calculation on the angles you get to start with. Then you probably will decide that you need a third reference if you want any kind of accuracy.

 

[jmilone]

Thanks again for the comments. Here are the latest developments:

We proposed another method and even wrote 7 pages regarding the subject, this is our idea:

Connect a PC to a Linx ES RF transmitter via USB with a Linx QS module. Authorize the sending of the pulse and stamp the time in a software resource. The transmitter sends the pulse at ~920Mhz with 1 bit of data. The target has a receiver also operating at 920 MHz. The data goes into the data_in of the receiver and connects directly to a transmitter. It transmits the data back to the original sender. The computer marks the time of arrival and calculates the elapsed time.

Constraints
- RF modules operate at 56000 bps
- USB max ~380 MB/s

Would the computer be able to evaluate these elapsed times given the fact they will be in the ns range? Our professor didn't think so, in fact he said we would need some extremely fast, digital, complex circuits to handle this kind of timing. Unfortunately for us, our proposal is due this Wed, so we don't know what to do. Does anyone think somehow it will work or is it too simple/good to be true?

Thanks

 

[Torben]

Seattle Robotics has an article on their site detailing a real-time laser rangefinder system. Essentially what it does is sweep a horizontal laser line vertically across the field of view of a camera; the image is then processed to isolate the received image of the laser line. Since the laser and the receiving camera are aimed the same direction but are mounted such that they are offset vertically from one another, parallax can then be used to determine the distance to various objects represented by portions of the reflected laser line.

Imagine you're holding a laser line generator at waist level and pointing it in the direction you're looking but angled upward slightly. The laser line will appear lower to your eye where it strikes objects which are nearer to you.

Anyway, the link to the article is here:

Not sure if that's an idea you can use or if it's really applicable to your project but it's another method which I don't think anybody's mentioned yet.


Good luck on the project,

Torben