Resolved behavioral problems with Avoider and Wanderer demo code by shortening sensor ranges

[grudolph]

In a previous post, I had described that when running the
TrackbotCommonV2 demo code, and WarmBodyHunterTest on a SunSPOT
Trackbot, that WarmBodyHunterTest ran as expected, but Avoider and
Wanderer did not run as expected. The latter two applications would
make the robot go forward and backward, or do nothing, in little
chaotic fits, with no apparent pattern.

Since the hardware seemed to be running fine (other applications run
fine), I surmised from watching the LEDs that the sensor ranges must
be too long, as readings were very chaotic even when the robot was
sitting still. Reflected signals must be giving false readings...
Once I figured out where the sensor ranges are being set, I compared
WarmBodyHunterTest settings to Avoider settings. Sure enough,
WarmBodyHunter sets the corner, side and cliff sensors to short range
(1), while Avoider sets the corner sensors to 3 (long), the side
sensors to medium (2), and the cliff sensors to short (1).

After setting the sensor ranges to short range in Avoider, both
Avoider and Wanderer demos work as expected in long, 9-foot wide
hallways.

[Bruce Boyes]

At 12:04 07/29/2008, you wrote:
>After setting the sensor ranges to short range in Avoider, both
>Avoider and Wanderer demos work as expected in long, 9-foot wide
>hallways.

Nice.

There's an App Note in progress about optimizing sensor range for a
given circumstance. Hopefully completed within a week.

In the meantime, Monitor mode of TrackBotCommonV2 shows the level of
ambient IR when you press SPOT button SW1.
to browse to the code:
https://trackbotcode.dev.java.net/source/browse/trackbotcode/trunk/TrackBotCommonV2/
checkout from SVN with this URL:
https://trackbotcode.dev.java.net/svn/trackbotcode/trunk/TrackBotCommonV2

best regards

Bruce

------- WWW.SYSTRONIX.COM ----------
Real embedded Java and much more
+1-801-534-1017 Salt Lake City, USA