Position sensing for small, indoor arenas

If you are really talking about an arena, have you researched the FoxTrax hockey puck tracking system?

Do you have to sense the position of one item or many?

Can you add hardware (a flashing IR LED, for instance)?

For a fixed installation and s slow-moving object, wireless webcams in each corner and a flashing IR LED on the object might work.

Mark Borgerson.

Look up an old 'experiment' more publicity stunt done years ago by Prof Warwick ar Reading University (UK).

He could do computer location of where he was basically by having a pet ID tag (RFID) located uner his skin, and his location to a room could be 'worked out' because of RFID readers near doorways.

Just wiring up all the RFID readers of suitable strength (within safety limits) and cabling/wireless back to control point.

--
Paul Carpenter          | paul@pcserviceselectronics.co.uk
    PC Services
 Timing Diagram Font
  GNU H8 - compiler & Renesas H8/H8S/H8 Tiny
 For those web sites you hate

Sorry, my bad. I meant arena as in "operating (physical) environment" -- not as in "stadium" :-/

Eventually, a large number. But, I'm looking to start out in a more limited specification and see what I can learn therefrom.

I can add anything I want -- on the sensed node or the "beacons". However, cost eventually becomes an issue. But, more significantly, "set up" has to drive the solution. E.g., I can't expect users to drag out a theodolite just to set up the "environment".

The LED would then have to be visible. E.g., you couldn't stick it in your pocket. (sorry, did I forget to explicitly mention I am tracking the location of the *user*...?)

I was thinking along the lines of a low power RF/AF beacon/receiver on the person and one or more beacons/receivers *placed* (i.e., not "positioned") in appropriate locations throughout the, er, "area" (avoiding "arena" :> ) that, IN CONCERT, can deduce the location of the user.

E.g., someone fresh out of school would set up beacons in precise locations, measure the distances and spatial relationships between them, calculate the absorbance of the walls and other objects in the signal's trajectory and *calculate* a POSITION.

I, OTOH, would be much more enamored of a solution whereby the user can walk into a "location" and inform the system/device "I am in the kitchen" or "I am in the north warehouse" or ... and walk around the boundaries (?) of that region while the system "watches". Repeat for other areas. Then, later, let the system make a judgement call based on actual incoming observations: "it *looks* like he's probably in region X".

This is a lot easier to set up. Again, note my comments regarding proximity, location, etc. instead of absolute

*position*.

This relies on maintaining state in the "sensor". I.e., you only know when the user has entered or exited from a particular region -- and, only if you have the ability to monitor every such egress.

If you "miss" a transition, you have no information of the user's location until (and *unless*) he walks by another reader.

As such, it's deployment in large areas becomes problematic unless you can *force* (i.e., *impose* an otherwise unnecesssary behavior on) the user to move within range of your sensors. E.g., you couldn't tell if he was standing in the northeast corner of the room or the southwest -- which, in the case of something like a warehouse, could be radically different areas.

I was thinking more along the lines of *finely* sensing Rx signal levels at "access points" (or their equivalent) as well as the levels at the *user* of the signals FROM different AP's (i.e., abandoning the idea of tying a user to a single AP) along with dynamically varying the power OUTPUT from those AP's (as well as the user's device) and looking at the resulting aggregate data.

E.g., as you decrease the Tx power from the user's device, you can get a better sense of his position/range wrt any particular AP. Likewise, in reverse (dropping the individual AP's outputs). Since this can happen quickly (assume the user is limited to foot travel), you can get lots of readings for each "location" (i.e., the user can be treated as stationary, even though he is really in motion).

...

Every system has features and 'benefits'..

You have last known valid for any system.

That depends on the tpe of warehose, if the warehouse is like an aircraft hanger yes for beyond range oif any sensor. If it is full of racking/shelving, then more than likely it would be easy to setup readers at end of each aisle, which most warehouses these days have other paraphenalia there anyway.

If you have a lot of racking shelving it cna make the radio methods more difficult.

Depending on size and layout of the areas involved could be problematioc, wireless works well for large open areas, may cause confusion for lots of small areas with different signals getting propagated differently depending on walls, moving furniture and numbers of people. May take longer to get baseline mapping of rooms.

Large area with aisles may work better with RFID at ends (possibly middle as well) or some form of IR system down each aisle.

IR of course has the additional problem of times of day being potentially flooded by sunlight. possibly hanger style warehouse main doors opened.

Even in a large open warehouse style, that has changing amount of 'wares' being 'housed' could cause you RF problems, as in the difference between an empty hanger and a hanger with two 747's parked. Even some materials that are piled up (salt, wheat, sand etc..) might potentially give problems with all methods for reflections shielding or even RF/magnetic pseudo 'short' because two much is stacked near the sensor.

You may have to employ multiple or different methods depending on circumstance.

--
Paul Carpenter          | paul@pcserviceselectronics.co.uk
    PC Services
 Timing Diagram Font
  GNU H8 - compiler & Renesas H8/H8S/H8 Tiny
 For those web sites you hate

Can the devices be 'trained'? They could perhaps store their own maps and 'report' where they are with a unique identifier.

Can you create a floor grid along the lines of the Kiva robots in the Zappos warehouse?

-to-conquer-warehouses/

Of course! Much prefered to enforcing "precision" on them by some rigid design/deployment criteria.

You're still faced with learning the "maps" (assuming the "devices" you mention are the "mobile units" and not the "stationary beacons [sic]".

Too much set-up involved.

Some scenarios to give you a flavor for how I would like to be able to deploy...

- imagine setting this up in your *home* (i.e., track which room you are presently in, etc.);

- imagine setting it up in a commercial retail establishment (i.e., track which "department" you/salesman are presently in)

- imagine setting it up in a warehouse environment (i.e., imagine which portion of the warehouse you are located in)

The last is probably the easiest to impose some arbitrary structure on. But, it is also quite large; you can't easily move/relocate existing storage areas; might be EMI/RFI "hostile";

*tends* to have folks who are focused on getting their job done (instead of cooperating with some "danged computer"); etc.

The household environment is the one I like to target when doing an initial evaluation of an idea. It's relatively small, few "targets" ("mobile units"), doesn't change that often (i.e., furniture and walls tend to stay where they are much moreso than a commercial retail establishment), etc. And, it is the *least* likely to tolerate lots of "set up" modifications ("No, you can't cut a hole in that wall! I don't CARE that you're going to patch it and paint it later...")

Commercial retail is probably the easiest to address as they are used to "remodeling" often (and typically have service areas in walls and ceilings already). OTOH, it can cost a considerable amount if you have to hire lots of tradesmen to do all the grunt work. :-/

Hi,

I think, its not so trivial at all. But what about some kind of modified VOR System?

The principle of the VOR is, that you have a signal which is AM-modulated by turning an antenna and an additional FM-modulation on the signal.

The phase between the FM-Signal and the AM-Signal at the receiver gives the angle where the beacon is, so with two beacons, you can triangulate the position of the receiver.

At an inhouse environement, you will get problems with reflections, but may be, such a system could be modified for inhouse use, e.g. if you combine an RF Transmitter beacon with an rotationg infrared lamp.

The RF-Transmitter defines the reference phase for the rotating infrared lamp. So, the receiver could see the lamp and the RF-Signal and from the time, when he gets the infrared signal and the phase of the FM-Modulated RF Signal, he could get the angle...

So, you would need only one strong RF Transmitter for the reference phase and several synchronized rotating lamps. The receiver could have several infrared detectors and the infrared signal could be modulated to decide, which led is seen by the detector.

best regards

Stefan DF9BI

it is difficult if you want the resolution too fine, but I know of one project that uses zigbee transceivers on packages and spread around in warehouses to track the packages.

Copenhagen airport is running a trial where you can "login" with your bluetooth phone and they will track you around and send and SMS when you need to start walking to the gate etc.

Aalborg Zoo used to have a system when you could get a bluetooth tag for your kids and you could then see where they were in case they got lost.

I don't know if the timing information is accurate enough to compute distances or if they just use RSSI to guess what fixed is closest

-Lasse

There has been a lot of research on "pseudo-satellite" systems.

Goggle has lots of hits on research and projects using this type of system.

Warning, its not cheap nor an exact science, yet.

h

I've looked at various indoor positioning systems before and all the RF-based systems that I've seen have a huge problem with one thing: walls. If your device is next to a wall and your absolute error is more than 4 inches, you've just placed your device in the next room. I've seen workarounds to this that involve keeping maps of doorways, staircases, etc. but it typically ends up in a big complicated mess that never really works very well. This is where ultrasonic will give you much better results and it avoids all the problems with those pesky things called walls. The system setup isn't really all that much, unless you consider hanging one or two sensors per room to be too much. There are commercial products out there (mostly for the health care industry) that use ultrasonic for indoor position determination and they're quite successful.

There is only one God :-)

Anyhow, the magic buzzword is "location tracking".

Ultrasound isn't going to work well inside buildings unless you are prepared to do a lot of hammer drilling and runing an infrastructure.

That pretty much leaves Bluetooth. You'd need devices scattered throughout the house that ping the device on the user. Possibly clandestinely, and some things such as certain cell phones, laptops and so on have Bluetooth built in. Of course that only works if turned on. Then the scattered motes could radio the read RSSI values from that device to your Orwellian command center, HAL the computation machine crunches those, then ... "hah, gotcha!" :-)

Correct, but the first thing would be to see if someone else has already dunnit, and these guy might be a good starting point:

--
Regards, Joerg

http://www.analogconsultants.com/

"gmail" domain blocked because of excessive spam.
Use another domain or send PM.

Why would you need to do any of that? These guys just mount one or two microphone modules in each room:

Do they claim to "have sight of" every package at all times? Or, do they just watch to see which gateways through which it may have passed? E.g., a local hospital tracks certain items in this manner -- but, only for certain "areas" (i.e., once the item has passed beyond their observation, they have no idea where it is and have to be prepared for it to REappear *anywhere*)

But, they don't need to actually *track* you in order to SMS, right? I.e., you could turn off the BT in your phone and there's no reason they can't still "text" you ahead of time (assuming they actually know how fast you walk and the location of your gate wrt your assumed present location; they could always fall back to a worst case of "let's give him

15 minutes advanced warning")

Now *that* shows promise! Though, without knowing the topography of the zoo, wouldn't there be areas where the child ("target") was out of range of *all* transceivers/beacons? So, in effect, all they could say would be "he/she was last tracked in this general area"?

Thanks! I'll look into it.

The obvious problem (ignoring cost, etc.) is that all it tells you is where you are wrt the beacons. You have no knowledge of the *beacons'* locations!

OTOH, as long as they don't move, you can train such a system by walking the perimeter of each "region of interest" after informing the system that "I am now located in region X" and let it map all subsequent pings to that region in its database.

Exactly (assuming 4.5" walls -- which isn't always a valid assumption). You can improve your resolution if you can control Tx power from each beacon "finely" (i.e., dial it down until you lose signal).

Note that optical and sonic schemes also have similar vulnerabilities as you have to consider the possibility of reflections (less likely with light but you can never be sure there isn't a reflective surface *somewhere* that lets light "around the corner").

You can reduce the problem areas by using redundancy -- but that makes set-up more exacting.

I don't see how that would eliminate the problem. Knowing where doors (and even walls!) are located doesn't help disambiguate the signals received.

You can get reflections from ultrasonic signals. You could shape the envelope of each "chirp" to differentiate between direct and reflected waves AT A SINGLE RECEIVER but, without "seeing" both signals, there is no way to tell if the signal you are receiving is the primary wavefront or a reflection.

I think you would have to operate in the high ultrasonic region (> 100KHz) to ensure (?) it wouldn't interfere with pets, etc. (in a domestic environment). I am unsure how prolonged (i.e. CONTINUOUS) exposure to this sort of radiation would affect humans, etc.

I'll call over to the local hospital and see how they track location. Of course, in that environment, I suspect "set-up" is typically EXPECTED to be cumbersome/expensive -- which might invalidate their sort of solution.

IIRC the attenuation of sound in the air is somewhat 3dB/m at 20kHz, and it increases as the square root of frequency. That renders high frequency ulrasound in the air impractical.

Vladimir Vassilevsky DSP and Mixed Signal Design Consultant

It can, but only if you're tracking faster than the object is moving. For example if the object is on the left side of a wall and hasn't gone anywhere near a door; even if your measured position is a few inches on the right side of the wall, you can be pretty sure that it is in fact still on the left side of the wall. This approach completely falls apart if you lose the signal for a few seconds.

Unless your transmitters have predefined timeslots for their chirps and you don't leave any large gaps without microphones. The setup of these systems is somewhat expensive; not in terms of parts or labor but in terms of planning.

The transmissions are not really continuous, they're short pulses at regular intervals and there's no significant risk to human hearing. They might sound like a faint ticking noise to some people, other people can't hear them at all. I have no idea how it would affect pets. It will interfere with acoustic glass-break sensors that are linked to burglar alarms but there is a fairly easy workaround for that.

Is your primary concern for setup the cost of equipment, the labor time, or the planning time? You can minimize two of the three.

I'm don't know, but I'd think they place the "base stations" in a grid smaller than the range of the transceivers so they always have at least one connection

I'm not sure what they want to use it for, but I'd guess it is just as much a way to figure out how people move around so they can improve and for the airlines, in case theres a missing passenger for boarding they know if he's around where to look for him and if he is on his way. probably advertising too

there's always the risk of being out of range, it's no different than a cell phone, but if the cells are small enough compared to the range it should work

-Lasse