SparkFun Forums 

Where electronics enthusiasts find answers.

Have questions about a SparkFun product or board? This is the place to be.
By hcawthorn
#154546
Hello.

For a while I've been wanting to build something that tracks my movement on the basketball court. Ideally I'd like to track my co-ordinates every second and then be able to download that information to a DB that I can then use to calculate how long I was in each location and where I went on the court.

There are very sophisticated systems that use colours on players jerseys and cameras to track movement but I'm looking for something a little more mobile.

I don't think GPS will work because the movement and speed I'll need the info passed will be too quick but was thinking about a class 1 blue tooth transmitter / receiver. Any one have any experience or advice on this? Any other suggestion?

I'd love to hear back!
By stevech
#154572
Long story short? Can't be done at 5m and less precision, affordably.

RF TDOA can work, but far too expensive. RF signal strength alone is way too ambiguous (inverse square law + line of sight needed)
By Mee_n_Mac
#154584
RTK GPS would have the precision needed but I don't think anyone's making a small enough receiver to be wearable. And then there's the cost and setup hassle. So GPS is out. You could wear some form of ultrasonic speaker and setup 3, more likely 4, microphones to triangulate your position. Wearing the speaker might be more of a burden than you care to bear during a game. You'll have to say how much of a hassle you can tolerate wrg that and setting up 4 poles with 4 mic's.
Lastly I suppose you could setup 2 webcams over the hoops and do some version of object tracking.
By stevech
#154618
RTK GPS needs
Outdoors
GPS Correction data feed
long baseline pair.

etc.


Bright IR diode on the moving person's beanie-cap; cameras in three places. Photogrammetry math up the wah-zoo.
To complicated and unwieldy for the OP's goa.
By jremington
#159653
Both companies use the building's existing wifi, with standard hardware. The key is that they walk through the entire building and make a map of the RSSI (signal strength) values, which is stored in a data base along with a building plan. Your location is then estimated from the RSSI and the data base.

RSSI is a very poor source of distance information and the accuracy of the position estimate would depend strongly on how many different wifi access points are present. Note that neither company states the positional accuracy, which I suspect is highly variable and at best, probably not much better than GPS (i.e. a few meters).
By nickersonm
#159720
Since you'll be in a defined area, and need fairly good precision, I also suggest visual tracking. If you have access to the ceiling, a single camera there would be sufficient, rather than two cameras in the X and Y planes. stevech's recommendation of an IR diode would make analysis much simpler: just look for the bright point, rather than trying to do object tracking.

You'd have to create a system to analyze the video, of course, but there may be tools out there you can adapt. I don't have any suggestions for this, although I've heard good things about OpenCV for object tracking.

You won't need any complicated math once you can find your position in the camera, if you place the cameras appropriately - either looking down, observing the XY plane, or on the X and Y axes to just get the Y and X positions. Calibration, if you need it, would just consist of you moving to the corners of the court.
By hcawthorn
#159874
I see what you're saying.

The one thing that I was thinking was that you wouldn't need to actually map out the court or play surface becuase 90% of the movement would be in the playing area which would allow a logic (TBD) to map out the court or rink which is just a rectangle shape. Also, ideally you don't need real mapping of traffic you just need a grid that acts as a heat map that shows how much time you spent a specific quadrant. Again not sure if that is even feasible but wanted to float that out there. Not really sure how the system registers where you are as opposed to the distance you are at.

At the end of the day if you could get how much time you spent on 1 half of the court (defense) vs the other half (offense) that would still give you some interesting insight.
jremington wrote:Both companies use the building's existing wifi, with standard hardware. The key is that they walk through the entire building and make a map of the RSSI (signal strength) values, which is stored in a data base along with a building plan. Your location is then estimated from the RSSI and the data base.

RSSI is a very poor source of distance information and the accuracy of the position estimate would depend strongly on how many different wifi access points are present. Note that neither company states the positional accuracy, which I suspect is highly variable and at best, probably not much better than GPS (i.e. a few meters).
By Chevelle
#159882
Hmmm. I'm thinking a combination of GPS and IMU. The GPS would get your rough location or at least a reference point every once and a while. With what I am sure would be very sophisticated software but a straight forward design using an IMU, the magnitometer, accelerometer, and gyro data can give you pretty precise information.

Your other option is to get/design a giant Kinect :)
By Hami
#159910
jremington wrote:Check out Sebastian Madgwick's video on footstep tracking, using just an IMU:

http://www.youtube.com/watch?v=6ijArKE8vKU
This will get much more complicated in sports.... . You have to think about new "zero velocity assumptions" (or any other boundary conditions) to get a result like that. Furthermore the estimated position probably will drift a couple of centimerters each minute even in a setup shown by Mr. Madgwick, i think.
Probably the exact absolute position is not necessary in your application? What are the outcome measures you want to determine? Just positions as you initially wrote? Is an absolute position needed?
By KeithB
#159916
Maybe you could wear the ultrasonic "pinger" from a sonar distance sensor and put 4 receivers around the court. Then you could measure the relative receive times to get location.
By hcawthorn
#160067
Don't really need specifics... just rough estimates. End outcome is to have an output that illustrates how long you were in quadrants of a court. Priorities of info need are:

1) How long you were in the offensive end vs defensive end
2) How much time you spent on either side of the goal so a player can see if they have bias to one side of the other

Even at that level of info it would be useful feedback.

Its not about specific but generalities that will let a player get a sense for what they are doing... really a heat map in that is based on 5 or 6 foot quadrants would be miles ahead for player feedback.
Hami wrote:
jremington wrote:Check out Sebastian Madgwick's video on footstep tracking, using just an IMU:

http://www.youtube.com/watch?v=6ijArKE8vKU
This will get much more complicated in sports.... . You have to think about new "zero velocity assumptions" (or any other boundary conditions) to get a result like that. Furthermore the estimated position probably will drift a couple of centimerters each minute even in a setup shown by Mr. Madgwick, i think.
Probably the exact absolute position is not necessary in your application? What are the outcome measures you want to determine? Just positions as you initially wrote? Is an absolute position needed?