simulating a robot which wil fully explore it's environment, and then
translate it's findings into a 2D representation of that environment,
rendered in OpenGL. i am doing reasearch of the background topics
currently. and will be in Miami on saturday to get to the library
there. that's about all for now
busy. i went to the miami university libraries on 9/20 (sat) and spent
considerable time there, but could not check out any books because i
did not have a community borrowers card. Nate and i returned on wed
9/24 and got a card and books. much reading was done throughout the
following week.
i learned about heirarchical control structures, subsumption control,
and various areas relating to the two. i read on reference model
architectures, planners, behavior generators, feedback/feedforward
control, sensory perception, world modeling and newsted ELFs. all of
these are discussed in detail in my paper.
considerable bit of resaearch, although my searching has not always
proved fruitful. very few relevent websites were found on a google
search of 'behavioral decomposition', but after broadening the search
to 'behavioral robotics', many more options were available.
i have sucessfully downloaded and set up Rossum's Simulator v0.5 on
the ACLs and Nate's computer at home. Rossums is a robotics simulator
which lets the user define the arena and many details of the
robot. having read most of the 70 pager manual, i can now define the
area with which the robot interacts and the robot contruction
including sensors and the wheel system. i have ordered several books
over ILL, and continue to do reasearch.
i'm not sure how to keep track of the position of my robot for maping
out the area, however, once the area is mapped, the robot will be able
to report it's position fairly accurately.
journal. anyway, last week i got some work done, although not as much
as i would like, due to the fact that we had a fairly complex
assignment due in OS last friday, among other things. i have created a
list of behaviors which my robot will use:
Drive forward
Avoid obstacles
Follow wall
Map making
Loop detection
Target sensing
the loop detection will determine if the robot has previously been at
it's current location, and if so will send the robot off in search of
another wall to follow. the Target sensing behavior will be used to
determine if the robot must solve a maze (detect a target at the
beginning), and when it has finished the maze (detect a target at the
end). these two targets will be different, so the robot will not
mistake the target at the beginning as a signal that it has finished.
behaviors in the RP1 simulator. i havn't had much difficulty in
programming these so far, and all the errors i have found have been
fairly simple, mostly interface errors. other things of note: my ILL
book came in last week, all about behavioral robotics. at last, i
will have a source which will focus specifically on my topic.
implementing a wall following behavior that actually works, which hasn't
been so succesful.
trouble with the simulator processing moting requests. currently, i
believe that the simulator has some queue of motion events that it
processes when it gets around to it. i'm going to extensively look
through the code for the demos to be sure i have the right syntax,
although i'm fairly confidant that i do. with the debugging options
turned on, i can see the simulator recieve the motion request, and it
then says 'Adding start-motion request(....). after this, there
should be a line that states 'sim start motion task[...]' which never
comes up for the rotation in my inside corner behavior. the backing
up now works all the time. I am quite confused, because i use the
same syntax for both calls and one works and the other doesn't. well,
back to work.