This week’s EYE ON NPI (video) is moving in the right direction – it’s u-blox’s XPLR-AOA Direction Finding and Indoor Positioning Explorer Kit! This NPI tries to solve a well-trod but still not-fully-solved technical challenge of how to perform 3D tracking of low cost/low power ‘tags’ indoors. Ironically, object tracking is something that humans and animals do very well – we can even track things that have hidden themselves from view! But for robotics, this is an incredibly hard problem.
Let’s talk about some of the ways that we can do object tracking now, to explain why something that sounds so simple has been a challenge for decades.
The way robotics do object tracking now is how the most simplistic organic vision systems work: by looking for a contrasting color or shape. This works best if something is vividly colored – like this Aibo robotic dog, tracking a round pink ball. It works great if the thing you’re trying to track happens to be round and pink, which is not many things on this planet. Also, it doesn’t work particularly well if the shape is hidden or obscured.
The next step up from basic shape tracking is machine learning vision systems that try to recognize objects by using layers of matrix calculations – what we would normally call OpenCV or TensorFlow. This is a little more flexible, but still requires good lighting, unobscured vision, and recognizable shapes.
If you don’t want to use a camera, and you happen to be outside, you can use GNSS/GPS. A board with a GPS module and a radio transceiver can pretty easily determine location and then relay it back to a central station. GNSS gives you up to 10 meters This is great for cars, people, boats – all sorts of large objects. But that 10 meter precisions makes it tough for smaller items and of course, GNSS does not work indoors. (If you do want better precision, you can get it using RTK – check out our EYE ON NPI from last year).
Traditionally, when working indoors to do tracking, folks have relied on a few different technologies. In particular, RSSI tracking is quite popular because it’s so cheap. Basically, radio signal strength falls geometrically with distance from antenna to antenna. While annoying when you’re far from your WiFi router, it does make for a basic way to determine distance. Get two or more fixed-point transmitters, and you can do basic triangulation on the RSSI to determine location. It doesn’t work great, and suffers from issues like multi-path bouncing and EMI affecting the signal, but it is very easy to implement and it’s pretty cheap. We see the technology used in ‘Beacons’ or ‘Tiles’ these days. You can get about 4 meters of accuracy here.
Folks can also use time-of-flight technology, which has slowly been making it to WiFi modules (and may also make it into BLE at some point – we’ll do EYE ON NPI on that technology when it filters into the market!)
As of Bluetooth 5.1, there’s a new Direction Finding capability built into the wireless specification.
Bluetooth direction finding makes it possible to determine the direction that radio signals travel from a mobile tag to one or several fixed anchor points. Using angle-of-arrival (AoA) technology, anchor points comprising an antenna array that is connected to a Bluetooth receiver can detect the direction, or angle, to the mobile tag, which transmits a Bluetooth signal. When a constellation of such multi-antenna anchors is deployed, AoA technology can be used to triangulate the precise location of a mobile device or tag.
Note that this is not distance measurements, it’s angular measurements. But, of course – if you have a few fixed antenna station locations it’s easy to convert a set of angles into a precise location! The angle calculations seem to give better accuracy than plain RSSI – 1 to 2 meters – and can be used for more than just location sensing. For example, in this demo from u-blox, a camera can follow a tag just with angular data since we don’t care how far away the target is, just that it is in frame.
Direction Finding is included in any u-blox/Nordic module that supports BLE 5.1, but to make it easy we recommend picking up a XPLR-AOA Direction Finding and Indoor Positioning Explorer Kit that has the antenna configuration needed and laid out. It’s a lot easier and faster than routing your own boards – the ‘tags’ are any BLE module and do not need special design considerations.
Lucky for us, the u-blox XPLR-AOA Direction Finding and Indoor Positioning Explorer Kit is in stock at Digi-Key right now, and is an excellent way to get started immediately with trying out the new technology. If you want a more advanced setup, with 4 fixed-point-nodes and 4 tags, sign up for the XLPR-AOA-2 kit.
See the kit on Digi-Key at https://www.digikey.com/short/q3dpn32p
See the u-blox video below: