Let's Make Robots!

Xbee conference notes

Gareth, David, Glen, Chris and I discussed Xbee radio modems. Here are some notes I took.

Watch part I  here.  Watch part II here.

TAG all your Xbee related content with the tag "xbee", please.

 

Xbee is a Radio modem system, aimed to replace wires on a serial network. Compared to BlueTooth, it has more features and applications suitable for robotics.

Two Xbee modems out of the box will communicate with each other. They are rigged to communicate on a certain "channel" which defaults to "zero" ("communicate with everyone"). Each Xbee can be configured with a unique "PAN" ID. This will prevent your Xbee from talking with your neighbour's Xbees. Choose the same PAN for all devices that you need to talk with each other.

Lady Ada has this to say about PAN ID: "By default all XBee's use PAN ID #3332. The ID is 4 bytes of hexadecimal and can range from 0000 to FFFF. XBees will only send/receive data to other modems on the same PAN." Compare the SSID in WiFi.

Xbee, like other 2.4 GHz radio data systems (BlueTooth, WiFi) operates in a band of the radio spectrum. Multiple frequencies are available and the device will select the most suitable (available) frequency for the situation.

On top of the standard "wire replacing" point-to-point communications, the Xbees (version 2 or newer) can also do "Meshed Networking". You need to work in "API mode" for this. Meshed networks forward data (in packets) from node to node until they reach their destination. They do not need a human-organized routing table for this, like Internet routers do.

The "get rid of all the cables" application of the Xbee has now been superseded by hobbyists like Gareth. He is trying to prove that you can also replace an expensive many-channel RC transmitter/receiver pair with a couple of Xbees. Most RC plane pilots already rely on Xbee for telemetry feeding back from the plane to the ground. But Gareth wants to see if Xbee would be suitable to actually control hist plane. From the ground to the plane. Potentially risking plane crashes. That is one good reason to build your own, inexpensive plane for experimentation.

Given the 2.4 GHz radio technology, it might be a very bad idea to trust an expensive (or otherwise valuable) vehicle to this radio signal. A dead zone can be hidden behind the smallest of obstacles. Even obstacles on board of your vehicle. It is always a good idea to place an external antenna on your robot, as far away as possible from any potential disturbances like your electronics or motors. Make it stick out or hang off.

But this concern is just as valid for commercial 2.4 GHz RC remote control sets (and for that matter, NASA's communication strategies between Mars and Earth). The difference with "our experiments" is the amount of testing and development put into the existing solutions. You get what you pay for. (Or at least, this is what we are hoping, as tax payers contributing to those Mars missions.)

Confidence in your system can be built up. For example, Patrick starts by building a ground vehicle before he lets his electronics control an air bourne unit. David suggests flying a regular RC plane through conventional controls. Attach a transmitting Xbee on it and evaluate signal quality on the ground based receiver. Or the other way around, for that matter.

Improving range or signal strength. Replace the half wavelength whip antenna (about 2.5 cm) with a bigger antenna. Maybe a dangling wire, maybe a commercial stick like the ones on your WiFi router.

Signal boosting using a "cantenna". Google it and you'll find examples for WiFi or BlueTooth. These would work as well for Xbee. Most gain is derived from noise cancellation, rather than boosting the actual signal.

The beauty of the small Xbee modules (with an SMD antenna on the board) is that they are small enough to place directly inside a metal can, antenna and all. Hot glue!

Alternative directional antennas include Yagi, helix and parabola dish. But designing and building those yourself is a pain. Every millimetre off spec will slash 2% of your wavelength.

Question from the shoutbox by Renkku: how about attaching an external antenna to the screw terminal (on some Xbee boards)? Specifically a "rubber duck" style thingey.
Answer: any 2.4 GHz rubber duck would probably work and improve your signal processing. But you want to connect it directly. No converting connectors please, or you will lose any gain or even some of the original signal.

[continuing my notes here]

Xbees as analog transmission lines. Xbees have some “left-over” pins. Some of those are can be used to read an analog value (voltage). The value can be transmitted (at preset time intervals) for the other Xbee to receive. The receiving Xbee can give out this value again as a voltage (or rather PWM signal).

Possible application: pulling a reset pin low on a microcontroller remotely, so that the digital link can then be used to program it. This enables “over the air reprogramming” of (for example) an Arduino. Gareth will be using this in his RC plane project “Xbee Files”: telemetry about battery voltage.

[This is where the Ustream crashed for a moment. The recorded parts I and II are approx. 3 minutes apart.]

Signal strength is measured and reported by an Xbee in API mode. The radio strength is then simply standard part of the data packet format.

Possible application for this is “the holy grail of robotics”: localization.

Different strategies:
1) directional antennas through “trivial math” and “simple trig”.

The directional antenna would have to be mounted on a panning platform (servo) with decent feedback about its angular position. Relative to North (when the antenna Is on a beacon), or relative to the nose of the vehicle (when on a vehicle).

Compare shielded IR transmitter / receiver pairs.

2) compare four (or three or five, whatever) signal strengths measured on the corners of your robot. See Gareth’s comments below. Or compare to a system the police are using to recover stolen cars (that are bugged with a hidden transmitter), called “LoJack”. (So steal one of their vehicles and hack their system, shouts CompleteTanker).

This could be changed to a system with transmitters on the four corners of the arena where one (or more) bots are roving around.


Tee-shirtable quotes from this show:

“We don’t want the math people to be part of the equation.”

"Stay with your mic and speak your mind!"

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.

Radio direction finding almost never concerns itself with distance primarily because usually either the transmitting station or the receiving station is capable of moving. since the station can move, the usual tactic is to take a bearing, draw a line across the map to that bearing, move to a new location, take a new bearing. Where the lines cross will give you the location of the transmitter, more or less. Because most mapping systems us Cartesian coordinates, trigonometry is used to get the distance. The algorithm is used to navigate on spheres(Earth) and flat grids with relative accuracy.

Issues arise with so called multipath, the tendency of a radio signal to follow more than one path between the transmitter and the receiver. Think of a radio signal bouncing off a water tower or perhaps a moving airplane. The same signal will arrive at different times at at different angles. This is a well known phenomena and is fairly easily dealt with by watching the way the signal behaves.

Multipath really requires a fair amount of distance separation to become a major factor. In the short distances we are talking about (under a mile) with relatively weak signals multipathing can be an issue but likely won't be that much. The real issue is that the question of whether or not the Xbees at the four corners of a robot, or arena, are sensitive enough to register a difference at all. Assuming that the answer is yeas, the system will work as that is the principle that phased array vertical operate in and these have been around for decades. It remains to be tested as to whether the resolution in the xbees are sufficient tot he task.

In the amateur radio word, there is a device that does this. It's called a DF Jr. which uses the the perceived signal difference between a set of antennas that are spaced 1/4 wavelength apart (This is a minimum not a fixed value).I've got one in my garage, designed for 144 mHz. NOt sure if anyone still makes them but...

Now there are several white papers on the subject of using xbees for localization available on the net (google xbee localization) done by a number of universities around the world.

Of course, one could build a real phased array antenna and feed it to a vanilla 2.5 gHz reciever and just do plain old vanilla radio direction finding.

commercial RC gear from XPS already uses xbee modules, so it seems proven already to me.