That's what they claim on their website, but you are right, I think Indoor Positioning System would be a better description of what Anyplace offers. <2m accuracy is good enough for me though, just wondering if there are any alternatives. And thanks for the Decawave suggestion, good to be aware of it but it is waaay too pricey as you mentioned
Your per unit cost of item tracked would be quite high, but there's a version of the client of https://www.internalpositioning.com/ that you can run on ESP32 boards. You'd need to figure out a power solution for the device too.
I've had success putting 18650 cells and then 3d printing cases for these: https://www.amazon.com/Makerfocus-Battery-Shield-Raspberry-A... Battery last long enough for peronal tracking around my house, but likely wouldn't be long enough to track the thing for multiple days in a row without a recharge.
I have had really good experience with a Microsoft Research project called Path.
The main usecase I had was navigating places like shopping malls without proper shop numbers or any addressing scheme.
Looks like in Cyrpus (.cy) there is second level domains depending on the site, and ac.cy is for academic and research institutions. so ucy.ac.cy is the university of Cyprus, cs is the computer science subdomain, and ap for the anyplace sub subdomain. It does look odd...
Very interesting. Indoor mapping is a great project to take on. I started mapping my house using Wifi Signal. By capturing signal strength, noise, and connected AP for each device for every second. I can manage to turn lights off and on as I enter certain rooms. I can also tell the "occupancy level" of each room and "Who's where".
I read the values right out from the AP and sends them to be stored in the DB. I had to build up a proper set of data (about a weeks worth) before I could go and "back fill" location. Uses some simple MA algos on the signal/noise. Working on "predicting" next location based on historical data.
I know the location of every WiFi device connected to my network. In general, I, and other occupants carry a device that has WiFi (Phone, watch, etc). Since I know the location of the device, I can infer/assume the location of the person.
Cool :) I have done my bachelor's project around wifi location fingerprinting back in the day. Took a year but with a bit of tweaks we did manage to calculate pretty accurate models of signal propagation by just having a wall model. The only downside at that time was the lack of development tools for Android (we had to use netbeans which was being horrifying experience). I guess it would be a lot easier to do the project with modern development environments + Flutter & Dart and using a bit more modern versions of OS :)
Really nice to see that this area is still active and being worked on!
I did the same for my bachelors but we basically pushed a cart with a wifi sampler around the building and built a machine learning model in matlab based on the signal strength to access points.
It would have worked waaaay better if we had some sort of surveying equipment to accurately capture each few steps of the cart.
Oh this is good, this does room level localisation by the looks of it. From what I can see, with Anyplace you have give it a building plan and then walk along the and sample the APs RSSI for each position, more data points give a better approximation of where you are
Am I correct in seeing that neither is available on Apple products? Is this because of some technical difference or is Apple putting up barriers or some sort of open source issue or something?
Apple doesn't give apps the ability to read raw rssi values from seen list of ssid(s). Also their WiFi Mac randomizing can cause issues depending on how you've got things setup.
Good for WiFi/IMU-based indoor (phone-based) position tracking, I believe. For more accuracy, especially for robotics, etc., Pozyx [1] claims resolution of "a few centimeters" (indoor rooms/outdoor) by using some custom trilateration circuitry, and then there are attempts to use VR (esp. Vive) methods [2], which are even better (albeit more limited range).
Has anyone used the in-store navigation feature in the Target app? It works on items that are in stock - it shows you the aisle number. Tap on that and it shows a store map (with the correct floor you are on) and the location of the item and your location. It’s scarily accurate. I’m not sure of the technology behind it.
I was curious about their technology too. From my brief research it works using Bluetooth beacons accessible for app via iBeacon API - https://developer.apple.com/ibeacon/
32 comments
[ 4.6 ms ] story [ 78.0 ms ] threadFor inventory you could use decawave, but it would be pricey.
I've had success putting 18650 cells and then 3d printing cases for these: https://www.amazon.com/Makerfocus-Battery-Shield-Raspberry-A... Battery last long enough for peronal tracking around my house, but likely wouldn't be long enough to track the thing for multiple days in a row without a recharge.
You can find it discussed here https://www.microsoft.com/en-us/research/blog/path-guide-new...
.ac.cy is the Cyprus equivalent of .edu in the US. Or, more similarly, of .ac.uk in the UK.
ucy is University of Cyprus.
cs is their Computer Science department.
ap is Anyplace, the name of the app.
Really nice to see that this area is still active and being worked on!
It would have worked waaaay better if we had some sort of surveying equipment to accurately capture each few steps of the cart.
I've been using that with quite a bit of success in automating my smart home as I move around it.
It is the Navigator Video on the far right column of the homepage.
Good for WiFi/IMU-based indoor (phone-based) position tracking, I believe. For more accuracy, especially for robotics, etc., Pozyx [1] claims resolution of "a few centimeters" (indoor rooms/outdoor) by using some custom trilateration circuitry, and then there are attempts to use VR (esp. Vive) methods [2], which are even better (albeit more limited range).
[1] https://www.pozyx.io/ [2] https://hackaday.com/2018/09/04/this-is-your-solution-for-op...
Edit: a word, and also this was on iOS.