INDONESIAN LOCAL AND ENVIRONMENTAL OBJECTS IMPLEMENTATION FOR AUTONOMOUS TRAM SIMULATION USING CARLA SIMULATOR

• Main Author: Mikha Erwansyah, Jeane
• Format: Final Project
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/76547
• Institution: Institut Teknologi Bandung

The research ”Development of an Autonomous System using Artificial Intelligence for Trams” focuses on developing an autonomous system for trams. This autonomous system’s development requires simulation to facilitate its implementation, and the simulation uses a ”software-in-the-loop” scheme with the CARLA simulator as the software tool. The simulation needs to be as close to the real world as possible to support the system’s development. This final project implements tram, local objects, and typical Indonesian environments in the CARLA simulator for autonomous tram simulation. After exploring the CARLA editor and its documentation, 3D model assets are created, those assets are then imported, the imported assets are edited, the edited assets are integrated into the simulation, and implementations are then validated. The implemented objects include trams, public transport vehicles (”angkot”), traditional bicycles (”sepeda onthel”), motorcycles, pedicabs (”becak”), train stations, traffic signs, and railway tracks. The results of the implementation are validated and evaluated to ensure that these assets or objects, particularly the vehicles, remain stable within the simulation and can follow both user control and autopilot control. The implementation results show that trams and angkots are stable as four-wheeled vehicles and can follow user control and autopilot control. On the other hand, traditional bicycles, motorcycles, and pedicabs remain stable as static objects and are not implemented as two-wheeled vehicles. Train stations, traffic signs, and railway tracks are stable as static objects as well. Sepeda onthel, motorcycles, and pedicabs are not implemented as two-wheeled vehicles because they are unstable during the simulation. This instability is caused by asset misconfiguration. The implemented assets are then exported, making them available for other autonomous system developers to use these assets.