Self Driving Car Models

This repository included various of applied models used in nowadays autonomous vehicles.

1- Introduction to Self-Driving Cars

kinematic bicycle model notebook

In this notebook, I implemented the kinematic bicycle model. The model accepts velocity and steering rate inputs and steps through the bicycle kinematic equations. We can provide a set of inputs to drive the bicycle in a certain trajectory.

The bicycle begins with zero initial conditions, has a maximum turning rate of 1.22 rad/s, a wheelbase length of 2m, and a length of 1.2m to its center of mass from the rear axle.

longitudinal vehicle model notebook

In this notebook, I implemented the forward longitudinal vehicle model. The model accepts throttle inputs and steps through the longitudinal dynamic equations. We can give a set of inputs that drives over a small road slope to test your model.

The input to the model is a throttle percentage in range [0,1] which provides torque to the engine and subsequently accelerates the vehicle for forward motion.

Controller2d

In this .py file, I implemented full logitudinal and lateral control of a simulated car in Carla software.

2- State Estimation and Localization for Self-Driving Cars

Extended Kalman Filter EKF

I implemented an Extended Kalman filter, which estimates the trajectory of a vehicle using odometry, range and bearing measurements.

Error state EKF

Implementation of ES-EKF to predict the vehicle position depending on Data from IMU, Lidar and GNSS sensors.

3- Visual Perception for Self-Driving Cars

Applying Stereo Depth to a Driving Scenario:

This notebook contains an inplementation to locate a car in a scene and estimate how far you are from it using a stereo camera model. This exercise gives a brief introduction to using Python to implement how to find the distance to collision with an obstacle.

Visual Odometry for Localization in Autonomous Driving

In this notebook, you will practice using Visual Features - Detection, Description and Matching to estimate an autonomous vehicle trajectory by images taken with a monocular camera set up on the vehicle.

Environment Perception for Self-Driving Cars

In this notebook, you can find:

How to use the output of semantic segmentation neural networks to implement drivable space estimation in 3D.
How to use the output of semantic segmentation neural networks to implement lane estimation.
How to use the output of semantic segmentation to filter errors in the output of 2D object detectors.
How to use the filtered 2D object detection results to determine how far obstacles are from the self-driving car.

4- Motion Planning for Self-Driving Cars

Occupancy Grid