Indoor Positioning System How-to
Description
This document describes an infra-red based indoor positioning system. The
system uses up to 8 cameras to provide position information on infra red
sources within their view. The cameras are static (no pan/zoom), so they
can't perform any tracking. But we believe that 8 cameras should be more than
adequate to provide full coverage of an average-sized classroom or laboratory.
We are experimenting to determine the coverage we can get and to determine
how many IR sources can be in the environment simultaneously without taxing
the processor capturing the data.
Contacts
To get more information or help with the system, contact
Anind Dey or
Joe Bayes or
Gregory Abowd.
The system came from the
Interactive Media Technology Center here at Georgia Tech. Our contact
person is Peter Presti, but
please ask one of the previously mentioned people any questions, before
contacting Peter.
System Components
Currently the system consists of:
- custom PC board and camera box
- data capture S/W (executables and C++ source) for Windows NT
- 4 Sony CCD cameras, filters, cables, power adaptors
- data processing S/W (executables and C++ source) for Unix
Status of System
Currently we only have the positioning system running with a single camera.
When we get multiple cameras working with this system, I'll update this
document. We have parts for 4 cameras, although only one is assembled.
System Access
The positioning system is in Room 102 of the College of Computing.
You'll have to see
either Anind Dey or
Joe Bayes for the password to the
computer.
Instructions
- Turn on the PC
- When given the option of the O/S to use, just hit RETURN
- When prompted for login, press CTRL-ALT-DEL
- Login using name and password obtained from Anind or Joe
- Once the system is done loading, start the capture.exe file located
in C:/Capture directory
- Once the program has loaded, under the View menu,
choose View Camera 1
- Connect camera cable to DC IN/SYNC connector of the camera
- Connect Video Out (red) to Greyscale Video Channel 1
on the camera input box. The channels are numbered starting
from 0, so you want to connect to the second connector on the
top.
- Connect VD In/Out (white) to Composite Sync Channel 1,
second connector on the bottom
- Plug camera power adaptor in
- Test system using IR emitter and battery or any IR-emitting device
(e.g. TV remote)
- Remember the IR emitter is a diode, so current will only
flow in one direction. You may have to flip the
emitter around to get it to work.
- Aim the IR source into the camera (start about 1 foot away)
- Should see a point/oval in the Camera 1 viewing window
- View the actual data by selecting View Packet under the
Debug menu. The data is in the following format:
x position, y position, camera number, percent intensity
- Change the view by selecting (checkmark) the Show Centroids
option in the Control Panel. I think the control is mapped
incorrectly. When the option is checkmarked, you only see a
single point representing the IR source. When the option is
blank, you see a red oval representing the IR source.
- Write the data to a port by selecting the Serve Data Over Network
in the Control Panel.
- On any SGI, run ~anind/CyberGuide/Tracker/Simul. This connects to
the positioning system computer and reads the positioning data.
It does a little processing (time averaging of data and some
other stuff - to be determined) and writes the data to
the screen in the following format:
x position, y position, 8 bit intensity, camera number
for each point/centroid the camera sees.
- To shut down the system:
- stop the Simul program running on the SGI
- exit the capture program (File menu)
- unplug the power adaptor
- shut down the PC using the Windows Start menubar
- power down the PC
Comments or suggestions? Email them to
fce-webmaster
Last updated: 4/21/97