Multiproc & Dynamics Compatibility

[manye31]

Note

Some of this might not be conflicting with the new dataset branch (im def not passive aggressively finishing this PR now :^)

Features:

• Dynamics datatypes (Racepak)
• Add flag: --fill_missing_stamps which gives optional behavior to pull timestamp from bag if no timestamp in topic (such as steering_angle)
• Add an optional optional arg to topics which are not always present e.g. racepak, RTK topics.
• Tried to add more helper functions to make script slightly less monolithic
• A crude multiprocessing functionality. See more below.
• Update viz with the multiprocessing functionality

Multiprocessing

Taking advantage of each topic's independence from one another, I distribute each topic to an independent process to process the data into KITTI format. Future work should also try to add multiple processes for a single, more computationally demanding topic like perception data and split up the frames to process.

I kept the ros2bag_2_kitti.py alone, and made a new ros2bag_2_kitti_multiproc.py script.

This multi-proc is relatively a drop in set of code with 2 primary functions: display_progress_monitor and process_topic_wrapper. The process_topic_wrapper is just a wrapper function around the generic bag reading code we already have, to be used with multiprocessing. The display_progress_monitor is just a nicer CLI UI to show proc progress.

Here's an example of the CLI output: