Tuesday, November 29, 2005

Status on Cricket

Currently I can get raw cricket readings via serial port (forwarded over TCP). So at the least, if we have a laptop with a serial port moving around, we will be set.

I do not have proper localization working, per se -- but I am simplifying what we want cricket to do for us (for now): Rather than localizing where someone is based on multiple crickets, we will do what our initial goal was which is just consider distance to a single cricket. Meaning that we will know our radius to a node independent of what other nodes are saying, and can calculate our position to an accruacy of 1cm.

This way we can determine when someone is close to a point and trigger an event.

One way to improve this is to increase the sampling frequency on the nodes so we can get a more continuous gradient and potentially get acceleration. Neat.

For now, I recorded a dump of the communication as a listener moves towards a node, and will use it as 'training' data while I work on a Java client that parses the stream and triggers events.