Debugging Process

CommonRoad-Search
1

Using my current implementation I fail the scenario ARG_Carcarana-8_4_T-1. Is there an established debugging process that I could follow? I tried using the tutorial notebooks to visualize what my algorithm is doing but I’m not shown anything. Any tips would be highly appreciated :slight_smile:

2

Hi Markus,

To debug, you can use batch_processing_sequential():

  1. In batch_processing_config.yaml, change “inputmode” from DEFAULT to SPECIFIC and add the scenario that you want to debug with in “scenarios_to_run”.
  2. Add breakpoints where necessary and run batch_processing_sequential.py

For visualisation, maybe this helps Is there any available method for visualization?

Best,
Ling

3

Thank you Ling for your response. I am trying to debug ARG_Carcarana-8_4_T-1 which is a pretty big scenario. Therefore the display_steps method seems to have problems with visualizing it even with the information from the link you provided. Is there any other way to follow the path my algorithm goes in the scenario?

4

Hi Markus,

Yes, you can still use display_steps to visualise your path. What you need to do is adjusting the plot parameters according to the scenario that you want to visualise in method plot_search_scenario( SMP.motion_planer.utility.py).

At the beginning you can comment out “renderer = MPRenderer(plot_limits=[55, 100, -2.5, 5.5])” and use “renderer = MPRenderer()” to find the range that you want to zoom in, then update the plot_limits accordingly.

For scenario ARG_Carcarana-8_4_T-1, the following setting works for me

plt.figure(figsize=(22.5, 4.5))
renderer = MPRenderer(plot_limits=[-40, 20, 50, 70])

Hope this helps.

Best,
Ling

3 Likes