Chapter 3: motion planning

Extract the complete zip into a folder, then open README.html. Keep the folder together. The reading retains all four appendices. A PDF copied out of this folder does not carry its companion files.

Some PDF viewers block local links or drop query parameters. Use the matching links below in README.html when that happens. The browser reading includes its mathematical renderer and fonts, so it does not need an internet connection.

Start the browser apps

From this extracted folder run:

python3 launch.py

On Windows use py launch.py. Open the address printed by the command if a browser does not open. This serves only this extracted folder on your computer. It needs Python 3 and no Python packages. Keep the terminal open while using the apps. Stop it with Ctrl+C. Use --port 8766 if port 8765 is already in use.

The sampling app also opens directly as a local HTML file. The optimization explorer, which has the Motion refinement tab, uses JavaScript modules and needs the local server. It is already built; you do not need Node.js or npm.

Link in the reading	File in this folder
figure generator	`lectures/04-D2-motion-planning/code/make_figures.py`
sampling reference	`modules/motion-planning/rrt_two_link.py`
edge demonstration	`modules/motion-planning/index.html?mode=edge`
cost comparison	`lectures/04-reading-motion-planning/code/cost_comparison.py`
section check	`lectures/04-reading-motion-planning/code/check_section2.py`
whole-link implementation	`modules/motion-planning/staged_two_link.py`
rai_nlp.py	`lectures/04-D2-motion-planning/code/rai_nlp.py`
stage4_rai.json	`modules/motion-planning/fixtures/stage4_rai.json`
native RAI program	`modules/motion-planning/stage4_komo_rai.py`
archived run records	`lectures/05-motion-planning/sampling-demo/evidence/rai-stage4/run4-before-defaults.txt`
seed-study record	`lectures/05-motion-planning/sampling-demo/evidence/rai-stage4/seed-run.txt`
Python reference	`modules/motion-planning/rai_bidir_rrt.py`
chapter check	`lectures/04-reading-motion-planning/code/check_continuation.py`
rrtstar_reference.py	`modules/motion-planning/rrtstar_reference.py`
refine_figures.py	`modules/motion-planning/refine_figures.py`
refine_rai.json	`modules/opt-explorer/fixtures/refine_rai.json`
sampling views	`modules/motion-planning/index.html`
optimization explorer	`modules/opt-explorer/index.html`

Python for the numerical examples

The appendices say to run their commands from the course repository root. In this package, run them from this folder instead. The supplied lectures/ and modules/ paths are part of the package. Create its environment on macOS/Linux:

python3 -m venv .venv
source .venv/bin/activate
python -m pip install -r requirements.txt

On Windows use py -m venv .venv, then .venv\Scripts\activate. After activation, use python in place of .venv/bin/python in the appendices’ Bash examples. Set PYTHONUTF8=1 if a Windows terminal cannot print a mathematical character. The numerical scripts need NumPy; figure generators also need Matplotlib. No robotics package is needed for the two-link calculations.

Each Bash block in the appendices is a separate example. Run each from this folder. The commands print their results. They write no figures unless you pass an option that asks for them. Comments and usage lines in some scripts name files of the course repository that are not in this folder; the commands in the reading do not need them.

Native RAI runs

The native RAI command in Appendix B.3 is optional. Its recorded results are included under modules/motion-planning/fixtures/ and modules/opt-explorer/fixtures/, and the NumPy scripts compare against them. The course ran it with robotic 0.4.1 on Linux x86_64. In the course repository, scripts/rai_run.sh starts that environment in a container. In this package, scripts/rai_run.sh runs the file with this folder’s .venv/bin/python, or with python3 when there is no .venv. To use it, install robotic==0.4.1 in that environment yourself. The package does not include RAI or a container image. Its RAI_NOVNC setting has no effect here.

Course records that are not included

Two links in the reading point to course records that are not part of this package. The module notes are written for the instructor, and their replay evidence is summarized in Appendix C.1. The Python reference modules/motion-planning/rai_bidir_rrt.py contains the replay code itself. The lecture whose figures Appendix A reuses is Deck 06; its slides on the course site carry the same figures.

License

See LICENSE.md for the course-material license and licenses/ for the notices of bundled third-party browser dependencies.