Page 313 - Kỷ yếu hội thảo khoa học lần thứ 12 - Công nghệ thông tin và Ứng dụng trong các lĩnh vực (CITA 2023)

Tran-Viet An and Vu-Anh-Quang Nguyen 297

which guarantees precise simulation of robot dynamics and interactions with their sur-
roundings. In addition, Gazebo can interact with ROS, a widely popular framework
used for building and managing robotic systems.
With Gazebo, developers have access to a comprehensive set of features, including
support for multiple physics engines, various sensors, and customizable robot models.
The platform boasts an extensive library including plugins that enhances its function-
ality, such as terrain generation, camera, and even weather simulation. Gazebo is capa-
ble of simulating complex scenarios involving multiple robots, obstacles, and environ-
mental factors, tool for testing and assessing robotic systems before deployment.Ga-
zebo is a flexible and robust platform used for developing and testing robotics and
ADVs.

Fig. 1. Gazebo designing interface

3 Experiments And Results

3.1 Methodology

We will start by selecting the scheduling algorithms used in ADVs, such as A*,
s, and EBAND algorithm, then build a simulation platform to test the
performance of these algorithms. The simulation platform will include a set of
representative scenarios to evaluate the algorithms' performance under different
conditions.To evaluate the performance of the three scheduling algorithms, we
implemented them in Python and embedded in Gazebo based ADV model and tested
them on a simulated environment. The simulated environment consisted of a map with
obstacles, and the ADV was required to navigate from the starting point to the
destination point. We used the following metrics to evaluate the performance of the
algorithms:

A. Path Length.
The path length is the total distance traveled by the ADV from the starting point to the
destination point. A shorter path length indicates a more efficient algorithm.

ISBN: 978-604-80-8083-9 CITA 2023