Project information
The goal of the project is to implement a
trajectory generation
scheme.The approach reduces the search space by considering the
optimization of specific cost functions which yield polynomial trajectories,
coupled with a feedback linearization control, to navigate in the Duckietown
environment, which consists of a simulator providing customizable city-like
scenarios with lanes and intersections. In this setting, the vehicle is
controlled using the information provided by the monocular camera on top
of the vehicle, responsible for recognizing the center lane that defines
the local Frenet frame. The simulations are in the Duckietown simulator
(based on OpenAI Gym) using Python.
