Re: Ultrasonic sensor interactions

In lugnet.robotics, Brian Davis wrote:

> only a few of us even have access to two NXT US sensors
> at this point.

   OK, and since I'm one of the few... I sent up a stationary NXT with an US
sensor, and a "target" NXT with an US sensor as well, pointed directly at each
other. Furthermore I had the "target" move back and forth to vary the range
slightly. I then looked at the output of the stationary US sensor in Excel for
six differently configured "targets": (A) US sensor configured as such, range 60
cm, (B) US sensor configured as such, range 20 cm, (C) US sensor present but
unplugged, (D) US sensor configured as light sensor with LED off, (E) US sensor
configured as light sensor with LED on, (F) US sensor configured as a touch
sensor. Results?

First, the baseline (C) resulted in 434 accurate readings. So it works (hey,
always good to check).

For 434 recorded US readings for each of the other situations, there were a
total of 544 mis-reads (a failure rate of 1 in 4... ouch). So it seems there is
a problem if both US sensors are just "left to their own devices".

Is there any variation depending on what the sensor is configured as? Well, no.
At least not any that I'd call statisticly significant. Here's the error rate:

(A) 44% (two US sensors cofigured as such, 60 cm range)
(B) 18% (as above, 20 cm range)
(D) 31% (configured as light sensor LED off, 30 cm)
(E) 18% (configured as light sensor LED on, 30 cm)
(F) 14% (configured as touch sensor, 30 cm)

Conclusion: at least I personally can't seem to shut of the US sensor from the
NXT-G language. It remains to be seen what can be done with NBC or replacement
firmwares, or even if LEGO addresses this in a future update or release (they
did a *lot* of that on the RCX).

-- 
Brian Davis