feat: add Learning-Based Accel/Brake Map Calibrator

[CristianGariboldi]

Description

This is the PR regarding the learning-based vehicle calibration for longitudinal dynamics already discussed and presented in the discussion thread and in the Planning and Control meeting.

Tests performed

This calibrator has been tested in simulation and real vehicles. We've been using the calibrator for a different range of vehicles in Pix Moving with high accuracy.

Additional materials here:
Pix_Longitudinal_Calibration.pdf
Calibration.pptx

Effects on system behavior

This new calibrator improves the previous one in different aspects:

1. real data are interpolated with a neural network, which is able to provide a very accurate and smooth map, and can capture the non-linearities of the system;

2. the whole calibration process is very user-friendly, fast and efficient;

3. you don't need a huge amount of data for training the neural network, making the data collection process faster;

4. the learning-based approach can be easily adapted to different kind of scenarios, and not solely on the longitudinal dynamic. In my repository for example, I also included the calibration for steering/parking scenarios, just by adding 1 input to the network;

5. It can be used with a diverse range of vehicles, with different dynamics, kinematics, engines and transmission systems;

6. It has been tested in simulation and with real vehicles. We now use it for calibrating our vehicles in Pix Moving.

Pre-review checklist for the PR author

The PR author must check the checkboxes below when creating the PR.

• I've confirmed the contribution guidelines.
• The PR follows the pull request guidelines.

In-review checklist for the PR reviewers

The PR reviewers must check the checkboxes below before approval.

• The PR follows the pull request guidelines.

Post-review checklist for the PR author

The PR author must check the checkboxes below before merging.

• There are no open discussions or they are tracked via tickets.

After all checkboxes are checked, anyone who has write access can merge the PR.

[TakaHoribe]

@CristianGariboldi Thank you for your contribution! This is so great! Would you fix some checks(DCO, pre-commit)? Then, we can run the CI test.

[CristianGariboldi]

@ahuazuipiaoliang here

[Owen-Liuyuxuan]

Thank you for your contribution, and the analytics are very informative!

I hope we could further organize the steering calibration file (rename or somehow re-factor them to one single file) if that is possible.

Thank you very much.

[CristianGariboldi]

Thank you for your contribution, and the analytics are very informative!

I hope we could further organize the steering calibration file (rename or somehow re-factor them to one single file) if that is possible.

Thank you very much.

Hi @Owen-Liuyuxuan , sorry for my late reply, busy times, I have some deadlines for my imminent graduation, as soon as I will be free, I will fix this PR.

When you say to re-factor the steering calibration files to one single file are you referring to the neural network scripts?

[Owen-Liuyuxuan]

@CristianGariboldi
Yeah sure. Because for now the network_scripts_*.py is quite confusing in naming and has quite a lot of repeated code. I hope this could be further organized.

[stale]

This pull request has been automatically marked as stale because it has not had recent activity.

[mitsudome-r]

@TakaHoribe @CristianGariboldi Could you share the latest status for this PR?

[mitsudome-r]

In my opinion, this could live under autoware_tools repository.

[CristianGariboldi]

Hi @mitsudome-r , unluckily, since I finished my internship months ago, I cannot have access anymore to the company repository in this PR. I am afraid I can't modify it nor submit a new version. Let me know if I can help.
@ahuazuipiaoliang FYI