GOALS SOFTWARE CODE FUNCTIONALITY
KIDNAPPING COZMO
Leah Attai, Casey Rhodes, Rachel Fuller

OUR GOALS

Our goals were to:

  1. Get the previous groups' code running on our machines and try to understand their code.
  2. Improve upon the previous groups' functionality by implementing Monte Carlo Localization after Cozmo is kidnapped.
  3. Document our code thoroughly.
  4. Create a website documenting our efforts.

Link to our documentation of efforts: Fourth Hour Notes

PROJECT REQUIREMENTS

The stuff you need to run this project.

CODE

The code you need to run this project.

MCLocalize.py
MotionUpdate.py
PictureTaking.py
Histogram.py
Stitching.py

Functionality

What went wrong, what went right

We were able to successfully recreate the previous group's accomplishments by getting Cozmo to generate a panorama and histogram. We were also able to modify their Monte Carlo Localization to generate a histogram that reflects initial guesses of Cozmo's location and where Cozmo thinks he is at after localization.

Blue represents initial randomly generated guesses. Orange represents where Cozmo thinks he is at after localization.


Improvements to be made:

  1. Although we are able to generate a histogram with where Cozmo thinks he is at after localization, the histogram should have spikes with high densities.
  2. We also needed to work on ensuring that the localization guesses wrap around the panorama. The way our code is right now, if we were to pick pixel 0 and make an image from going half the width left and right, then we would end up with half a black image. Ideally, it should fully wrap around.
  3. Our motion model should update the new pixels with the pixel difference from turning 10 degrees when Cozmo is localizing. There should also be no upper/lower bound.
  4. It would also be good to have something that prints out where Cozmo thinks he is at during localization to ensure it is accurate.