Position Tracking Devices

[DonRotolo]

It would be helpful to know the precision requirements you are thinking of.
I am not aware of any off-the-shelf gizmo that does what you want; there probably is such a thing, you need to spend a few hours on Google.

The easiest thing to track would be a light source. Radio is directional, but not nearly as directional in such a small space as light. Also, tracking RF precisely is much more difficult than tracking a light source.

There are linear CCD sensors available - think of a camera that has only one row of pixels. An inexpensive one might have 128 pixels (1 x 128), so you can detect 1/128 of the field of view (which needs to be managed with lenses). Using two such sensors, you can triangulate (in real time) the location where the lines intersect - but at low precision at larger distances. The solution to that would be to use multiple sensors, perhaps six (one at each end plus two on eqach side - but this might have a blind spot in each corner) or even a few dozen.

The player would wear a belt (a necklace seems dangerous to me) with a small battery pack and a few LEDs of a certain color - other colors need to be filtered out of the sensor's incoming light. (IR is probably the easiest).

Pick the sensors that report the highest brightness every (pick a time period, maybe 50 milliseconds), use a lookup table to determine the abcolute location based on the pixel value reporting highest brightness, and you have your X,Y player location.

Write back if you want to get into details, including devices.

Don

[KenWittlief]

reading this thread again, it has just occurred to me, a camcorder looking down from the ceiling will track where the player is on the court all the time, all by itself

all you have to do is look at the live video screen, or play the tape back, and the player himself will move across your monitor exactly the same way he moved around the basketball court. "There he is, there he is...."

This almost sounds silly when presented this way, but it brings out an issue with your original question. You have jumped into the middle of the engineering design cycle with this thread, and asked for help with the HOW part: How to design a tracking system

but you never told us the WHY part (the problem you are trying to solve, which will explain WHY this is needed) and the WHAT part (what the system is required to do to solve that problem).

The more you explain the WHY and the WHAT, the better we will be able to brainstorm the HOW.

An off the shelf video camera will track a basketball player on the court through the whole game, with a graphical representation of his X,Y position. Is that all you need? I get the feeling there must be more to the problem you are trying to solve.

[Andrew Blair]

Haha, I do think Ken is right. But, if you want to do some serious coding (It looks like the other solutions are starting to get that way), you could buy an infrared camera and track the blobs, assuming you intend to see how far they run, how faast they are, acceleration, etc.. Thirty thousand lines of code later, and you can glean whatever you want.

But seriously, if you want to know how far a player runs in a game, how about a cheap pedometer? Might seem like a cop-out, but is probably more attainable before the basketball season is over.

[dhitchco]

Congrats to ken for making all of think "outside the box" and what is the actual original problem statement.

What is the actual "need" for tracking people on a basketball court?

Can the observer be a human just 'watching' the overhead video and drawing a player's track on the TV screen?

Can the observer be a PC doing vector triangualtion between two radio receivers?

What's the need?

Maybe the players can just "call out" their locations as the move.

Anyways, here's a really cool company that uses something called ultrawideband radio transmitters and receivers on peopel (belt-worn) inside building to track their exact location at any time:
http://www.ubisense.net/bfora/system...sttitle.xsl/14

[KenWittlief]

Herding cats! I had a professor, senior year at SUNY Buffalo, Dr Schmidt, who shared his Golden Rules of Engineering with us from time to time.

One was: Never design a system unless the requirements are clearly specified. Its better to quit your job than to try to design something if you dont know what its suppose to do.

If you try anyway, in the end the chances that it will happen to do what the customer really wanted are near zero. The odds are the project will be a complete failure, and you will lose your job, and you mind in the process.

Better off only losing your job.

You gotta know the WHY and the WHAT before you can invent the HOW.

[BRAVESaj25bd8]

My group and I need to track the movement of a player on a home basketball court, meaning it will be outside. We plan to return the ball to the player wherever he or she may be on the court. We do not need to know the distance to the player as we feel any home court is small enough to just aim the ball back to the player and let the momentum of the ball take it to the player on a bounce or two. This is only for shots that will be made and it is only for one player games. The intent is to retun made shots back to the player so he or she does not get out of rhythm when they are trying to practice and improve their shooting skills. My group and I came to the conclusion that we would need a position tracking device that works up to around 30 or 40 feet.

On an aside, thank you to all who have posted, your ideas are giving us a lot of options to look into.

[seanwitte]

The CMUCam and a fluorescent traffic vest would be a workable solution.

[Alan Anderson]

Maybe you don't actually need to know where the player is. Perhaps detecting the direction the ball came from and returning it in that direction will be good enough. That only requires looking at a relatively small area near the net, rather than at the entire court.

[yongkimleng]

Originally Posted by seanwitte

The CMUCam and a fluorescent traffic vest would be a workable solution.

True! This could detect angle from a fixed position (probably the ball launcher).

Perhaps if you want a true X-Y detection system of the range of a half court, you could place an ultrasonic sensor on a servo to perform a sweeping motion. Allow the sensor to perform a full scan before doing calculations and launching the ball.. which could take 2 ~ 3 seconds, which should be acceptable

[seanwitte]

Originally Posted by yongkimleng

True! This could detect angle from a fixed position (probably the ball launcher).

Perhaps if you want a true X-Y detection system of the range of a half court, you could place an ultrasonic sensor on a servo to perform a sweeping motion. Allow the sensor to perform a full scan before doing calculations and launching the ball.. which could take 2 ~ 3 seconds, which should be acceptable

I was thinking more along the lines of a rotating ball collector with an angled "ball return ramp" placed under the net. The camera would be mounted on the rotating base, keeping the return ramp pointed at the vest the whole time. Then you don't care about the angle and you're always returning the ball to where the person IS not where the shot came from. It would also be portable, just put it under the net and turn it on. The algorithm for the camera would be to keep the blob centered in the field of view.

[KenWittlief]

A couple years back the team I was on used the laser diode sensor with the retro tape. You are right about the 2 to3 seconds to acquire the target (using an ultra sound range finder), but once you have detected the player on the field you can have the servo sweep back and forth on a smaller angle, maybe 20 or 30 degrees or so. That way you get quicker updates on position (instead of sweeping the whole court, when you already know where the player is.

You could also do a little bit of target-lead prediction - from consecutive readings you would know which way the player is moving, and toss the ball infront of them.

I am curious about the rest of the system. How will you retrieve balls? will there be a ball hopper that the player has to fill? When the ball is shot will the system indicate a ball is being fired so the player can anticipate it?

Also, if the system is going to retreive loose balls, how will you stop it from trying to pick up the family pet, or the players little brother when he runs across the court?

[KenWittlief]

Originally Posted by BRAVESaj25bd8

We plan to return the ball to the player wherever he or she may be on the court. We do not need to know the distance to the player as we feel any home court is small enough to just aim the ball back to the player and let the momentum of the ball take it to the player on a bounce or two. This is only for shots that will be made and it is only for one player games. The intent is to retun made shots back to the player so he or she does not get out of rhythm when they are trying to practice and improve their shooting skills.

Ok, now we are getting somewhere. The big WHY is: This system is needed so a person practicing on the court alone can keep his pace up without having to run after the ball or needing someone to toss balls to him

and the WHAT is: you want to design a system that can place a ball back into play, preferably projecting the ball in the direction of the player.

Very cool. Will this system always be in one place? for example, will it retrieve balls that go through the hoop, or will it be able to go after missed shots and bring the ball back to the edge of the court?

If it will always be in one place (attached to the goal pole for example) then you dont really need to know where the player is on the court (X and Y location) - you only need to know the angle from the center of the court to the player, so you can shoot the ball onto the court along that line.

This narrows down the requirements for HOW to design the system significantly!

You could use something like the retro reflective sensors and retro reflective tape (that was in the KOP a few years back, not sure if its still being used) The scanner has a laser diode that is modulated. The retro reflective tape sends light back the way it came, like the material used on license plates.

You could have the sensor either on a servo that sweeps back and forth across the court, or have it spin 360 degrees. The player would wear a headband with the reflective tape - whenever the sensor is pointing at the player the sensor will trigger.

Read the servo angle at that instant and you have your direction.