more specifically, suppose I have a bunch of small widgets that can be put together into larger blocks, e.g. using slots. We will label them w1, w2 … w100 etc. I want to be able to make a “radar” that will pick up connections between the widgets, e.g. w2 is connected to w5, (w3, w4) and (w3, w15), and so on. “Connection” is commutative and corresponds to having the two widgets joined; maybe this could be implemented as an electrical connection, but I am agnostic here. Ideally this radar should be a gadget that is already available off-the-shelf with a usb connection, or at least something with potential to become such. Widgets ought to be really cheap, i.e. we don’t want to put a dedicated microprocessor into each one or anything like that.
Presumably the different widgets could contain unique RFIDs, and of course there are already off-the-shelf devices that pick up presence of RFIDs. But is there a straightforward way to implement and detect connections between RFID labeled items? Or is RFIDs a dead end path and really there is a better, well-known way of handling something like this?
so you’re using the term radar metaphorically. my knowledge of basic electricity says the way to detect a closed connection is to install a detector along the circuit of connection. it could be a resistometer or a simple LED. if you have 100 widgets, then you have a marix of connections numbering 100 x 100. will a console of 10,000 LEDs be too difficult to wire onto your system?
Are you thinking about something like XBee or ZigBee units? they send/receive radio signals to allow communication between objects. They have fairly low power units. They also have units designed to work in a mesh configuration, which sounds like what you are describing.
“radar” is a metaphor for a device that can identify the connection graph at any given time. The context of the problem is such that it is strongly preferable for this to be “wireless” with a line of sight of several meters. Notice that the widgets can be assembled into various disconnected blocks i.e. the entire connection graph is essentially always not “connected”.
A much less preferable alternative implementation would be a device that can read off the connection graph for a single block (this graph is of course a connected one) by coming into contact with it, i.e. the user would “inspect” a single block by touching it with the gadget.
yeah, on second thought maybe RFID is false analogy and false hope. Perhaps it should be closer to an RLC circuit composed of two parts, one for each of the 2 widgets that are plugged in together. So for 100 widgets we would need C(100, 2) = 5050 possible oscillators with recognizable frequencies that can be picked up by the scanner. This all has to work only within a few meters, but I don’t really know what is the granularity of civilian permitted radio frequencies for things like WiFi, radio controlled toys etc.
OTOH maybe oscillators could be distinguished both by frequency and amplitude, to reduce consumption of frequency spectrum? Or would a scanner that picks up both frequency and amplitude be a lot more expensive than just the frequency one?
It might be helpful if you describe what you are trying to achieve. By partitially designing the solution you are making it hard to figure out what the problem is that you are trying to solve.
If the widgets are yours, then you just need to have each one transmit what it is connected to. This doesn’t need RFID necessarily - just give each widget an id, and include in the joining interface a way to decide which widget transmits the ids of all members of the block. Transmission could be infra-red over the range you are talking. You could simplify the problem by insisting that each unified set of widgets has a transmit widget included in it - otherwise every widget would need a transmitter.
If the widgets aren’t yours, then visual inspection is as good a way as any. As specified, you could solve the problem by just looking at them, couldn’t you? This is why we need to know the actual problem.
Dioptre, the problem I am trying to solve has been outlined above in mathematically abstract terms (connection graph etc). I don’t feel like describing the potential invention more concretely.
Yes the widgets are “mine” and are prefabricated part of the (hypothetical) system. So they can contain any electronics or whatnot, as long as cost per widget is low. When you say that the widget should figure out what it is connected to and transmit it, so is that going to be an active transmitter or a passive one (sort of like my proposed oscillator that responds to signal from the scanner)? If an active one, would this kind of gadget be cheap? How long would the (very small, and possibly impossible/hard to change due to logic of the domain) battery last?
Would $0.3-$1 per gadget be a reasonable price range for the transmitters you are describing?
You’re not interested in tracking “widgets”. You’re interested in tracking “connections” (scare quotes since we’re dealing with secret hypotheticals here, not actual things).
It sounds like your connections might not even be electrical or RF. e.g. two lego blocks snapped together are connected by physical mating. Multiple kites might be connected by strings, etc.
if the connections are the object of interest, put the smarts in the connections or focus the detection on the connections.
What about an optical solution, like the balls they use in motion capturing? Maybe you could print unique IDs on each widget (or maybe a 4-color scheme like in world maps, or maybe reflective tape combined with preprogrammed constraints to separate the points [such as if three tape points are within 5’, they are connected; otherwise they are not – dependent upon how your widgets are actually clustered, etc.]) etc.?
Then you could track 'em with a system of cameras, or even one camera that you manually re-angle.
LSLGuy, well, duh! Obviously I want to track the physical connections, I said as much. The electrical connections that I am proposing are just a way to make the physical connections “visible” to monitoring equipment.
Reply, no, optical tracking will not work. In many cases you would not even see some of the widgets when looking “from the outside” because they would be hidden by other widgets or unrelated objects.
You could probably do something like this using a small microcontroller like an RF PIC. Your connections between blocks would have to be electrically conductive, but you could use those as communication channels between the microcontrollers. The microcontrollers themselves could develop and maintain the node and connection list. It might be easiest to have one of the microcontrollers designate itself as the “master” and have it maintain the list as well as respond to communication requests.
It wouldn’t be off the shelf. You’d have to create a circuit board using one of these microcontrollers. Cost might be an issue. The microcontrollers themselves are only a few bucks each, but you’ll also have other components as well as the circuit board and assembly costs to factor into it.