Week 5 Submission

This week you will submit your updated code for Lab 8 and (optionally) Lab 9. Preferably push all your changes to your Git repository and submit instructions on how to clone and run your code on Canvas similar to previous lab submissions. File uploads are still allowed since some students have expressed challenges regarding the use of Git; however, we encourage everyone to learn using Git so that we can eliminate this option in the future and have a uniform approach.

Your submission for Lab 8 will be evaluated using the autograder.py provided with the lab. We will evaluate two capabilities:

1. The filter’s estimation can converge to correct value within a reasonable number of iterations.
2. The filter can accurately track the robot’s motion.

We will 5000 particles and treat their average as the filter’s estimation. The particles will be initialized randomly in the space. We define the estimation as correct if the translational error between ground truth and filter’s estimation is smaller than 1.0 unit (grid unit), and the rotational error between ground truth and filter’s estimation is smaller than 10 degrees.

In autograder.py the robot will follow a circular trajectory. We provide several example circles in autograder.py for your testing, but in the final grading we will use another 5 different circles and the score will be the average between these five tests. So make sure you test several different cases to ensure the reliability.

The grading is split into two stages, the total score is the sum of two stages:

1. (3 points) We first let the robot run 100 time steps to let the filter find global optimal estimation. If the filter gives correct estimation in 50 steps, you get full credit for this part. If you spend more than 50 steps to get correct estimation, 0.01 point is deducted for each additional step required.
2. (3 points) We let the robot run another 100 time steps to test stability of the filter. The score will be calculated as based on the percentage of approximately correct pose estimations.

Please see Canvas for a rubric with further details on grading.

Lab 9 will not be graded but code submitted for Lab 9 will be visually inspected and the TA will take note of anyone who demonstrates localization on the Cozmo robot in the provided physical arenas. Completing optional parts of weekly labs will be taken into consideration for your participation score.