IMPLEMENTATION OF CONDITIONAL IMITATION LEARNING ALGORITHM ON AUTONOMOUS DRIVING SYSTEM

• Main Author: Changgraini, Nelson
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/63001
• Institution: Institut Teknologi Bandung

The process of building Indonesia’s new capital city in East Kalimantan uses a concept called “smart metropolis”. One way to realize this development concept is by presenting selfdriving vehicles. Therefore, researches and developments of self-driving vehicle in Indonesia needs to be carried out. One way that can be done as a first step in self-driving vehicle’s research is by implementing imitation learning algorithm on the self-driving vehicle system. Therefore, this research aims to implement imitation learning algorithm on a 1/10-scale car. This research began by simulating imitation learning on a self-driving car simulator. Then, the research continued by modifying an available 1/10-scale radio-controlled car into a self-driving car system. This system is built by integrating several hardware and software in the form of programming scripts that were written in Python. This research also modified an existing learning model to be a simpler one. Finally, the self-driving car that has been built was tested to drive autonomously on a self-made track to derive conclusions of this research. This implementation was successfully carried out by testing the self-driving car on the created track. The car was tested to move autonomously with several left and right turnings according to command input. As a result, the self-driving car manages to stay on the track from the start point to the end point of the travel.