A hybrid collision avoidance system combining Deep Reinforcement Learning with Model Predictive Control, designed for autonomous vehicles in CARLA to navigate scenarios with stationary obstacles.

This repository highlights the integration of neural network-based control with PID and MPC approaches in the AirSim simulator to enhance steering inputs for autonomous vehicles. Employing imitation learning and a hybrid neural network architecture, the project aims to create a robust and unbiased model for improved autonomous vehicle control.

Derivation & simualtion of the the dynamics of a rigid-body double pendulum with a moving base, control torques, and an external wind force .