Assignment 2 information

• Support code:

  • assign2.zip — assignment support code

    Although there are a bunch of files here, the only one that you'll need to write code in is localization.cpp. There are some comments in this file with some further instructions.

  • You will need the Simple Vector Library (SVL) for matrix computations and operations needed in this assignment. We've created slightly updated versions of this library:
    • svl-1.5-win32.zip
    • svl-1.5-unix-tar.gz — to install on my Linux system, I did a make linux_RH. You can then type make and then make install.

      In my case (on the CS department FreeBSD systems), after doing make linux_RH, I edited the top-level Makefile to edit the DEST variable (because I'm not installing to /usr/local, and I also had to edit makefiles/config.mf to make sure it was compiling and linking with the same version of gcc (3.4) that I'm using to compile my programs (because the default in the CS environment is 2.95).

    You can find the original distribution of the code here. I believe the only changes were to change one line of Mat.cpp to replace the abs function with the fabs function. The windows version has an updated project file.

  • Although all of you already have this installed, I'll just mention that the support code also depends on dolt-lite, which is available on the Assignment 1 page.

Written questions (part II)

1. Convince us that your program is working correctly by submitting test cases (problem files, command files, etc.) that you used to test your code. I suggest you test your code by creating situations where you can observe each part of your program, for example, tests where you can see the sensing update reducing uncertainty in the robot pose is different directions.

   Name the problem files for these tests: test1.txt, test2.txt, etc. Turn in these problem files (and associated command and robot files) electronically with your source code.

   Turn in on your written work a brief description of these tests, including what they are designed to test and the results you got from your program.

2. Create a test input that illustrates "filter divergence". Name the problem file divergence.txt and turn it in (along with associated command and robot files) electronically with your source code.

   Briefly describy, in your written work, what happens in this test: what should happen and why this test shows filter divergence.

3. For calculating confidence limits, one of the points in Section 4 of the assignment handout says to take the upper left 2x2 submatrix of Px, i.e., the variance of x, the variance of y, and the covariance of x and y. The lower right element in Px, the variance of theta, is used for calculating confidence bounds on the orientation of the robot.

   This means that the covariance of x and theta, and the covariance of y and theta (the element in row 1, column 3 (also row 3 column 1) and in row 2 column 3 (also row 3 column 2)) are not being used. In general, these numbers are not zero.

   What does this mean about the position and orientation and their uncertainty? Why are we ignoring these elements?

Turning in your code and other files

• Create a directory (or rename your assignment 2 directory) with your RCS user ID.

• In this directory, put all the files there that are necessary to compile your code. Most important will be the localization.cpp file, but also include the other files (*.cpp, *.h, Makefile, etc.) that were included in the assignment 2 support code. Also include the test problem files for the questions above.

• Do not include object files, an executable, the svl library, or the dolt-lite library.

• Create a zip file or a gzipped tar file of your directory — this should unpack to a directory with your RCS user ID and all your files should be in that directory. The size of your zip file should probably be less than 30K.

• Email your archive file to irastaff@cs.rpi.edu.