A rossum of robots

Continuing from yesterday’s post, let’s explore this idea of calculating one’s location by analyzing the time delay from two separately located sound sources.

It would be easy to apply this principle to a swarm of robot vehicles. In this scheme, a stationary transmitter containing two physically separated speakers sends out periodic ultrasonic pulses. Each robot vehicle contains a clock, so it knows what time each pulse was supposed to have originated. By measuring the delay in receiving each of the two sounds, the robot can calculate its own location.

We can also do this in 3D, with a swarm of flying robot vehicles. In that case we would need a transmitter containing three speakers, physically arranged into a triangle, but the principle would be the same.

In any case, we can broadcast digital instructions saying where we want each robot to be. When any given robot is told where it should be, it will know which way it needs to go. This allows us to centralize the planning for the swarm’s behavior into a single computer, rather than requiring each robot to be smart on its own.

It would be fun to think of uses for such a swarm of robots — whether for artistic or practical purposes. But what would you call such a flock? A flock of crows is called a “murder” (hence the title of that work at the Park Avenue Armory). Bowing to literary precedent, perhaps it should be called a rossum of robots.

3 Responses to “A rossum of robots”

  1. J. Peterson says:

    This brings to mind the creepy UPenn quadcopters.


    Unfortunately, audio doesn’t work with buzzing propeller blades, so they have to resort to external cameras : (

  2. Stephan Ahonen says:

    Rather than periodic pings, how about taking a page from GPS and using pseudorandom noise? CDMA principles work about as well with audio as they do with radio waves.

    I’m a live sound engineer and use an audio analysis program called Smaart to tune sound systems… It can find the distance from a loudspeaker to my measurement mic with a resolution of a couple inches or so, even with arbitrary audio. I think if you blast Gold codes on your “locator” loudspeakers, you should be able to get enough SNR between them at the microphones that you can achieve real-time location as fast as you can keep doing laplace and fourier transforms… Provided you’re not moving the mics so fast that you get doppler effect.

  3. […] Thanks to Stephan Ahonen for suggesting the use of CDMA for sending homing info to a rossum of robots. […]

Leave a Reply