Can't find the results now but from what I remember through put was close to maximum theoretical. latency was not great.
The way it was measured was by writing a small code for the CM5 that receives a packet from the serial (zigbee/bluetooth) interface and sends back something. very simple. on the PC there was a code that measured the time it takes from sending till receiving the packet back. that done a few hundred times, using different packet sizes gives a good estimation of effective latency and its sensitivity-to-packet-size.
The motivation was to see if some kind of closed loop control could be done from the PC over bluetooth. ie: robot sends all servo positions and all sensor values to the PC which does some number crunching and sends back servo commands. If this can be done at 200 times/sec then you can do anything with the robot from the PC. if only 5times/sec then you can do things like path planning and relying on gaits pre-programmed in the CM5.Statistics: Posted by limor — Fri May 29, 2009 10:13 am
]]>