Problem Set 3, part 2

Download the pset files in a .zip

Download here a .zip with the notebook and a couple helper files

The second part of pset 3 focuses on robust pose estimation from point clouds. The auto-graded tests use synthetic data. We also ask you to try the method described in the notebook on real data collected from your lab session. If you'd like you can take a quick look at the static .html version of part 2.

Pset 3, part 2 is due on Gradescope Monday 10/07/19 by 10pm. The drake tag for this pset is drake-20190929.

---

Introduction

There are four types of files in this problem set.

Jupyter notebooks ( .ipynb).

The notebooks are usually the first thing you open after downloading the zip file. They contains questions you'll need to answer and code snippets to generate visualizations and test your code. The notebooks are meant as a sandbox for you to experiment with different ideas, and you do not need to submit them.

A (few) .py file(s).

These files contain the implementation of most functions used inside the notebooks. The bulk of the pset is about completing/modifying certain parts of these files, as per the instructions given in the notebook. You will need to submit some of these files for grading.

A test script with a name like this: test_pset_3.py.

This is the same test script on Gradescope that determines how many points you score. The test script is provided to you so that you can run the tests locally and quickly get feedback/confidence on how well you do. Do not modify the test script.

Model/resource files.

.sdf, .obj and .npy files fall under this category. They are used to represent robots and point clouds.

---

How to do this Problem Set

First make sure you've gone through the install instructions for Drake + Docker.

To run the notebook, do this. (Make sure to have the notebook script in your terminal's working directory, and specify the path to the pset3_2 unzipped folder)

./docker_run_notebook.sh drake-20190929 /path/to/pset3_2

For example if you have the pset3_2 directory in the same parent directory as the docker_run_notebook.sh script, you can run:

./docker_run_notebook.sh drake-20190929 pset3_2

---

How to submit this Problem Set

We will use Gradescope for collecting and grading your problem sets, since it allows us to auto grade your code against our software tests, as well as look at your code and other text answers. If you did not already do this, make an account on Gradescope, and add 6.881 with the class code emailed to you. Although all course content is open, we only do grading for officially enrolled students who are taking this class for credit. Please contact the class staff if you are taking the class for credit but have not received the access code.

Please note that you need to make two submissions for the problem set.

The submissions are:

1. Autograded submission
   • Upload your planar_alignment.py and ransac.py to "Pset 3, part 2, Code submission" on Gradescope.
   • In our testing, Gradescope will give you a grade within 5 minutes. You can resubmit as many times as you'd like before the deadline.
2. Lab report submission
   • In the last part of pset3_robust_preception.ipynb, you are asked to write a short report on running your code against real data. Include your answer in a PDF file and submit it to "Problem Set3, part 2, lab report" on Gradescope.

Please carefully follow the directions for each submission.

Good luck and have fun!

---