PolyU URIS

Project 1

Title: Decision-making and Control for Connected Autonomous Vehicles

Member: ZHANG Yuanqi  (AF, YEAR-2)

Abstract: Unmanned ground vehicles (UGVs) have made significant progress in recent years and have enormous potential to improve traffic efficiency and safety. However, decision-making remains a major challenge for UGVs, especially in complex traffic scenarios. Therefore, an end-to-end strategy has been proposed to overcome these limitations, which integrates perception, decision-making, and control processes into an end-to-end system, directly learning driving decisions from raw sensor data, and avoiding manual design and alignment of multiple modules in traditional methods. In addition, a novel and efficient multi-agent reinforcement learning (MARL) method is proposed, which models each vehicle as an agent to achieve exploration, communication, and collaboration, and uses MARL as the baseline algorithm. A human-assisted RL framework is proposed to address the issues of local optima and security in reinforcement learning (RL). This method also allows real-time intervention in actual deployment to enhance the runtime security of RL-based UGVs. RL-based UGV navigation strategies will be trained and evaluated through these technical solutions in simulated and real environments.

Project 1

Title: Development of an Intelligent Trash Pick-up Robot

Member: Mr WANG Zian (EEE, YEAR-3), Mr CHEN Xianzhe (EEE, YEAR-3), Mr WU Hengyu  (EEE, YEAR-3)

Motivation: Cleaning trash on streets is a boring job, and mobile sweeping robots are a good choice to replace human workers. However, the existing sweeping robots are mainly restricted to flat ground, and they blindly clean designated areas, which is inefficient generally.

Objective: This project aims at developing a mobile robot that has high terrain adaptability yet can carry the modules of auto navigation, computer version and unmanned control, to perform automatic trash classification and collection.

Project 3

Title: Applying Evolutionary Game Theory to Analyze Autonomous Bus Control System Algorithms

Member: Mr FENG Zhanpeng (ISE, YEAR-4) 

Motivation: The development of autonomous vehicles, including autonomous buses, has been a primary focus of research and development in recent years. Autonomous buses have the potential to significantly improve the efficiency, safety, and sustainability of public transportation systems. To fully realize these benefits, it is necessary to develop practical algorithms that govern the behavior of these vehicles.

Objective: The project will focus on the application of evolutionary game theory to the analysis of algorithms used in autonomous bus control systems, investigating how various strategies employed by autonomous buses affect the overall performance, efficiency, and safety of the transportation system.