Abstract
With the increasing complexity of ship simulations, advanced artificial intelligence (AI) approaches become crucial for enhancing the efficiency and safety of evacuation processes. This thesis aims to develop an AI system based on Reinforcement Learning (RL) to manage evacuations within a ship simulation environment, focusing on Non-Player Characters (NPCs) behavior as virtual agents guided towards evacuation points.The research methodology involves several stages. First, the creation of a realistic ship simulation environment is carried out to provide a context that approximates real conditions. Next, a Reinforcement Learning (RL) model is constructed, and the model is trained. During the training phase, two methods to be compared, namely Proximal Policy Optimization (PPO) and Soft Actor-Critic (SAC), will be used to obtain optimal policies in evacuations. After the training stage, system implementation is conducted, followed by performance evaluation within the context of NPCs case studies. It is expected that the results of this research will not only provide a deeper understanding of artificial intelligence in the context of ship evacuations but also have significant practical applications, particularly in simulating NPCs behavior in ship emergency situationsThis research aims to enhance understanding of AI applications in ship evacuations, utilizing cumulative reward to reflect agent strategy effectiveness, and episode length as a measure of the time required to complete evacuations. The findings are expected to provide new insights into the development of more adaptive and responsive ship evacuation systems.
Original language | English |
---|---|
Title of host publication | 2024 International Seminar on Intelligent Technology and Its Applications |
Subtitle of host publication | Collaborative Innovation: A Bridging from Academia to Industry towards Sustainable Strategic Partnership, ISITIA 2024 - Proceeding |
Publisher | Institute of Electrical and Electronics Engineers Inc. |
Pages | 226-231 |
Number of pages | 6 |
Edition | 2024 |
ISBN (Electronic) | 9798350378573 |
DOIs | |
Publication status | Published - 2024 |
Event | 25th International Seminar on Intelligent Technology and Its Applications, ISITIA 2024 - Hybrid, Mataram, Indonesia Duration: 10 Jul 2024 → 12 Jul 2024 |
Conference
Conference | 25th International Seminar on Intelligent Technology and Its Applications, ISITIA 2024 |
---|---|
Country/Territory | Indonesia |
City | Hybrid, Mataram |
Period | 10/07/24 → 12/07/24 |
Keywords
- NPCs
- Proximal Policy Optimization
- Reinforcement Learning
- Soft Actor-Critic