zhizhenqin / quantifying-safety-of-learning-based-self-driving-control-using-almost-barrier-functions

Quantifying-Safety-of-Learning-based-Self-Driving-Control-Using-Almost-Barrier-Functions

This repository contains the parameters of Kinematic and Dynamic models used in the paper (revised) Quantifying Safety of Learning-based Self-Driving Control Using Almost-Barrier Functions.

---

Vehicle Dynamics Models

Kinematic Model	Dynamic Model

The figures above illustrate the kinematic and dynamic vehicle models. The only parameter for the Kinematic model is the wheelbase of the vehicle, which we use $L=2.9$ meters.

For Dynamic model, the parameters are shown in the table below:

Parameter	Description	Value
$L$	Wheelbase (m)	4.35
$m$	Mass (kg)	1600
$I_z$	Yaw inertia (kg·m^2)	7500
$l_f$	Center of gravity to the front axle (m)	2.175
$l_r$	Center of gravity to the rear axle (m)	2.175
$c_f$	Coefficient for front tire cornering stiffness linear approximation (N/rad)	2e4
$c_r$	Coefficient for rear tire cornering stiffness linear approximation (N/rad)	2e4

The implementations of the dynamics are based on [1], [2] and [3]. We are working to get the full codebase of the barrier function training and verification soon.

[1] Snider JM. Automatic Steering Methods for Autonomous Automobile Path Tracking. 2009.

[2] Sakai A, Ingram D, Dinius J, Chawla K, Raffin A, Paques A. PythonRobotics: a Python code collection of robotics algorithms. Published online 2018. doi: https://doi.org/10.48550/arXiv.1808.10703

[3] Dong C. PathTrackingBicycle. GitHub repository. https://github.com/Derekabc/PathTrackingBicycle. 2020.