playground

CLARA Notebooks

A summary of multiple notebooks which are used for debugging and plotting.

Get the data from our 1718 season

It can be downloaded here. The default directory until now is playgrounds/logging-data-1718.

Directory

• scripts
  • all python scripts which can be used for multiple plotting and debugging purposes
  • plot-cluster-information-csv.py
    • create a valid cppdebuginfo.py file with multiple numpy arrays of our clustered data
    • read the ground truth from a csv file and plot them vs each other
  • plot-t3-tp-log.py
    • plot any of our t3-tp loggings
  • plot-cones-path-t3-tp-log.py
    • plot the blue/yellow cones, the planned path and the driven trajectory based on 4 csv files
  • plot-xy-positions.py
    • used to plot the different positions based on the changed yawrate
  • plot-skidpad-path.py
    • plots the middle line and the real driven line based on the skipa repository debug skipa_debug.log
  • plot-yaw-comparisson.py
    • compares the different yaw rates by calculating the mean and std, and plotting them against each other
  • plot-velocity-comparisson.py
    • compares the different velocities (Correvit and the calculated from visual odometry)
  • plot-any-logging.py
    • a slightly more generic plotting approach to plot any csv values next to each other
• notebooks
  • ipython notebooks which are used for iterative development of algorithms and intermediate results
  • gmm-code-examples.ipynb
    • we show how a gaussian mixture model works with examples
  • fuse_darknet_etas_logging.ipynb
    • example for our old file format. This is probably outdated but can be easily adapted for your formats. The key value is the filename of the image
  • plot-yaw-pos-comparisson.ipynb
    • a whole notebook for yaw position comparissons. Obviously dependent on the debugging information created with tests/clara_csv_replay

How to run notebooks

Install anaconda.

> conda create --name clara python=3.6
> source activate clara
> conda install numpy matplotlib ipython
> conda install -c anaconda ipywidgets
> conda install --channel=conda-forge nb_conda_kernels
> jupyter nbextension enable --py widgetsnbextension
> pip install mpld3 

Start the notebooks like this, to allow big logs if you need them:

> source activate clara
> ipython notebook --NotebookApp.iopub_data_rate_limit=10000000000