| Details: |
|
Date：
|
01/08/2025 (Friday)
|
|
Time：
|
13:00
|
|
Venue：
|
LT2, Wui Chi Building, Main Campus
|
|
Student：
|
Yuhan Chen
|
|
Topic：
|
Algorithms and Solutions for Intelligent Transport
|
|
Abstract： |
Intelligent transportation has been a hot and noticed area in recent years. The research aims to address the growing challenges of transport using advanced technologies such as artificial intelligence, machine learning and blockchain. It contributes to smart mobility through the development of autonomous algorithms for ground transportation and smart information system solutions in the air service industry. This thesis explores the use of deep reinforcement learning and 3D-LiDAR in autonomous driving. Deep reinforcement learning is capable of learning directly from high-dimensional data through trial and error and offers significant advantages in developing autonomous systems capable of dealing with real-world uncertainty. The use of DQN enables agents to efficiently learn discrete actions in controlled environments, whereas the DDPG algorithm extends this capability to continuous action spaces, allowing for smoother control of vehicle dynamics. By capturing local and global features in the 3D environment, this study explores various point cloud preprocessing algorithms to further enhance the perception module, resulting in more accurate object recognition and environment understanding. Experimental results from both simulated and real-world environments demonstrate the effectiveness of using deep reinforcement learning in the continuous control of self-driving vehicles. To address the challenge of secure and efficient data sharing within an intelligent transport system, this thesis also proposes an information system based on Distributed Ledger Technology (DLT). This system aims to ensure the security, integrity and efficient management of real-time data, which is essential for coordination and communication between different entities in an intelligent transport network. By leveraging the decentralised and tamper-proof nature of DLT, the system ensures the integrity of the data, thereby enhancing trust between stakeholders and facilitating the integration of the various component participants into the wider transport infrastructure. |
|
Enquiry：
|
fca@mpu.edu.mo
|
|
|---|
