© 2001 jukal at cyberian dot org
Extreme parental supervision
What do we need parents for? Let the technology mind what your children do!
 
I have a 3-year old kid. He is fast. Very fast. We live by the lake. Sometimes you think about the worst-case scenario - what if he doesn't believe our instructions and jumps into the water. In today's shiny happy world full of technology, cant I just say where my kid can go and let the machines take care?
 
So the target is to create a system which fills the following requirements with the given priority:
  1. monitors the position of one or more childs in realtime
  2. is able to act in case it recognizes that the child is about to exceed the virtual fences set as the valid playgorund.
  3. is able to react based on the activity performed by the individual being monitored.
  4. is safe
  5. is shock-resistant, water-resistant & dust-resistant
  6. is portable or "user embeddable" :)
  7. functions minimun of 7 days on a single load.
  8. has a cheap per day cost
 
I see multiple technical solutions to propose for this. Let's discuss them, try to choose the best solution and discuss the chosen solution in more detail. But first, let's describe the environment and the needs in more practical way.
 

Environment

Our home is in semi rural area, in a town near a bigger city. The place could be considered rather safe for children, but there are some details which may pose a risk, such as: the nearby lake, the well, the swing, a larger rock, some bushes with non eatable berries. These for example. So, if you are for example a system security specialist, you could see these as threats. What we need to do is to find the counter measures to reduce the risk posed by these threats.
 
They key is to react before the threat activates. So for example, if the kid is approaching the lake, if he is spending too much time around those dangerous berries or is playing with the well - we need to know in time. And this is what we need our technology for.

Techniques

  • Induction loop

    We spread wire around the zones which are considered safe, or near the entrances to unsafe area. Then we plant a let's say 2 kg iron load in the bag of the children. Our induction loop recognizes when the exits on enters a given area. [link: Understanding Basic Metal Detection]
     
    This solution has multiple disadvantages: it requires physical installation, which means sweat for the dad. It requires physics from the child, which means sweat for the kid. It is not bullet proof, it is not capable of deciding whether it was the kid or something else metallic such as an UFO flying by, it is also hard to utilize this technology to act based on activity.

    We reject this solution.
     
  • Heat based / thermal motion detection
     
    This solution is self explanationary. We place sensors around the area, and detect the kid based on the emitted thermal energy [link: How thermal ir imagers work].

    Let's go straight to disadvantages: requires installation and field maintenance. We have no way of knowing whether it was our kid, neighbour's kid, a rabbit or an elk. Also, the amount of heat emitted by the child varies depending on the amount and type of cloths weared by the kid. It could be possible to create a thermal profile for each individual, but if we take the amount of variations into account, it is hard to imagine it working. This system is hard to calibrate.
     
    We reject this solution.
     
  • Microwave based motion detection
     
    This solution is like an advanced version of the thermal (and other infrared based techniques) but in addition we could go through some obstacles and thus the need for sensors would be smaller. [link: Short Range Outdoor Microwave Transceiver]
     
    Problem: We still cannot recognize which of the targets is the correct one: is it a mouse or our kid.
     
  • Vision/image based tracking
     
    We place a digital camera somewhere above the area to be monitored. Either we plant a high pole on the ground or use similar technique as used with weather balloons to make the camera float above the area (tied to a rope). [link: How much weight can a weather balloon carry?]
     
    In theory this is a rather good solution as one camera can build a relative good overview of the area. However, there is still the problems of objects blocking the vision. Also the processing of the image data requires quite heavy computing power. Also weather changes for example pose a problem to this solution. And - would you like to have a damn camera sitting above your home all the time?
     
  • GPS - Global Positioning System
     
    In case someone does not yet know what GPS is, read it here [link: What is GPS?]. In this solution we let the child to carry a GPS receiver with a link to a central system. The receiver sends new coordinates of the child for example every 5 seconds. This data is collected by the server, which has a map of the area and defined hotspots (areas that cause some action to be taken).
     
    This solution is good, let's take it.

GPS-based child monitoring - do it yourself

Now that we have chosen the best tracking technology for this task, it is time to define the rest of the playground. How do we:
  1. get the location of the child
  2. transfer the location to a server
  3. analyse the location and finally
  4. act based on the current and previously stored information about the location
A GPS receiver with a RF transmitter solves our first two problems. The child watch from Digital Angel corporation is the perfect solution for us. It's a watch and a GPS receiver packed in a normal wrist watch, shock and water resistant and operates with standard watch battery. Child watch from Digital Angel
 
We also need some co-operation with the hardware provider to get the specifications of the system - they don't seem to be completely public at the moment.
 
Then we go to our Linux box, connect the RF receiver using serial cable. We read the specifications of the receiver and write drivers for it.
 
Once we are capable of communicating with the receiver we code a program which stores the position data in a database. See, we need to have access also to data from 1, 5, 10 and 30 minutes ago to compute the activity profile of the child being monitored. It also doesn't harm to store data from previous days - this can be used as a seed for fuzzy logic when analysing potential behavious profiles.

Then what we need to do is to map the environment. Place the threats where they are located: the stones, the lake and other hazards - we also need to map the topology of the environment.

Ok. Then comes the trickiest part. We need to write software which takes the current and previous positions of the child and analyzes whether the current actions and location of the child are acceptable. If not we need to trigger an alert based on the severity of the current actions: if the action does not pose immediate danger to life, such as the kid has been eating too many apples ( we have plenty ), we can just send an SMS short message to the parents phone. If, instead the hazard is extremely dangerous we can send a little electrical shock to the kid.

So, now our software is ready and we can think if there was any sense in this at all: In my opinion there was not - it is just so easy to think about the possibilities enabled by the technologies and forget that we are human beings.
 
<< Back to Amazon Discoveries
 
 
Copyright © 3/Sep/2001 Jussi Kallioniemi [ jukal at cyberian.org ]
The oddities presented here (that are not copyrighted or patented by someone else, ofcourse) are released under the GNU Free Documentation License.
 
 
Cyborg in ear: "nice orgy bar".
"Bingo!", Ear cry.