Your answers are at least as good as my questions :-).
On Mar 16, 2007, at 4:17 AM, Douglas S. Blank wrote:
but I'm unsure what this means (the robots I'm using are ones I
inherited from Lisa's asst, there doesn't appear to be any bumper
sensors on these robots). I would have thought hits referred to
locations that the robot is currently touching something. (My lack of
familiarity w/real robots here, in addition to conflation in terms,
is likely the problem; nonetheless I hope others might benefit from
these basic questions, which is why I'm sending them to the list.)
Yes, that text is more confusing than helpful. Here, "hit" is
suggestive
of the ray that extends from the range sensor and then intersects
with an
object in the world. So, the (x,y,z) tuple is the location (in local
coordinates) where the sensor detected an object.
So what's the difference between hit and distance (or value)?
Given hit values, is distance uniquely determined?
I am very glad to hear stall is a local measurement. Perhaps in real
life it would be implemented by noticing wheel slippage somehow (via
an increased motor current)? Actually, the more I think about this,
the harder it seems to determine on a real robot when a stall is
occurring, because even with feedback like quadrature, that's wrt the
axles, not motion on the ground. Almost seems like you'd need a
camera and image analysis to know for sure that a robot isn't moving.
I realize that, as these questions dig deeper, it would be valuable
for me to know more about the code's internals---something I was
hoping to avoid, at least until the end of my course. I plan to baby-
step start this effort by changing the trace/color locally.
Regarding the wiki and need for a central place to put answers to
these kinds of questions, I agree something is needed. The Pyro
online stuff is good, but things are still scattered. For that
reason, I tended towards writing my own subsumption Pyro-using
instructions for the students, and I then pointed them off to useful
pages like the Sensors page. But even with that, I've not been able
to find answers to basics like "what are the units of move's rotation
and translation". Thus, for instance, in my subsumption lab, I say
that these values can change from -1 to 1, but that if you'd like to
know what meaning these values have (i.e. their units), then issue a
command like robot.move(0,1) and estimate how long it takes for 1
complete revolution. Later I tried robot.move(0,15) and was very
surprised to see it turning even more quickly. I could have sworn I
found some online text about the move range being 0 to 1.
On another note, wrt subsumption layers, if I issue a move in a
particular update function, and then I "try it out" with the step
button, it only moves for a brief period of time and then stops.
Given the stuff in my prior paragraph, I'd have expected it to move
until the next step command was issued that declared move should be
something different. Is it b/c, with the button press, the simulator
goes ahead and models the fact that a layer in the associated run-via-
Run-button would only run for a fixed period of time before the robot
had to reexamine its sensors? Also, is it the case that a subsumption
layer is considered as possible for being active in the current time
step only if SOME move command (it could even be 0,0) is issued by
its update function?
Thanks again!
--b
_______________________________________________
Pyro-users mailing list
[email protected]
http://emergent.brynmawr.edu/mailman/listinfo/pyro-users
