SCALE@NTU Research Webinar Apr 2022

This research webinar on Teams is organized by Singtel Cognitive and Artificial Intelligence Lab for Enterprises (SCALE@NTU) to share the research work in the Corp Lab. For registration, please visit:

https://wis.ntu.edu.sg/pls/webexe88/REGISTER_NTU.REGISTER?EVENT_ID=OA22040717410586

Talk 1: Robust Road Network Representation Learning: When Traffic Patterns Meet Traveling Semantics

Road network is a fundamental yet indispensable component in transportation systems, which plays an important role in various downstream applications. We propose a robust road network representation learning framework called Toast, which comes to be a cornerstone to boost the performance of numerous demanding transport planning tasks. We propose two modules to learn representations which can capture multi-faceted characteristics of road networks. Specifically, we extend skip-gram to enable the model awareness of traffic patterns by incorporating an auxiliary traffic context prediction objective, and utilize trajectory data to extract traveling semantics. We design a benchmark containing four typical transport planning tasks to evaluate the effectiveness of Toast. Comprehensive experiments verify that the proposed method consistently outperforms the state-of-the-art baselines across all tasks. 

Speaker: Chen Yile, PhD Student, SCALE@NTU

Yile Chen received his bachelor's degree from Wuhan University, China in 2018. Since then, he has been a PhD student in Singtel Cognitive and Artificial Intelligence Lab for Enterprises (SCALE@NTU), Nanyang Technological University, Singapore. His research interests include data mining and machine learning, especially their applications in spatio-temporal analytics.

Talk 2: Multi-Agent Pickup and Delivery with Individual Deadlines 

We study the multi-agent pickup and delivery problem with task deadlines, where a team of agents execute a batch of tasks with individual deadlines to maximize the number of tasks completed by their deadlines. Existing approaches to multi-agent pickup and delivery typically address task assignment and path planning separately. We take an integrated approach that assigns and plans one task at a time taking into account the agent states resulting from all the previous task assignments and path planning. We define metrics to effectively determine which task is most worth assignment next and which agent ought to execute a given task, and propose a priority-based framework for joint task assignment and path planning. We leverage the bounding and pruning techniques in the proposed framework to greatly improve computational efficiency. We also refine the dummy path method for collision-free path planning. The effectiveness of the framework is validated by extensive experiments.

Speaker: Liu Yihao, PhD Student, SCALE@NTU

Yihao Liu received his B.Eng. degree in electrical and computer engineering from Shanghai Jiao Tong University, China, in 2019. He is currently a Ph.D. student in Singtel Cognitive and Artificial Intelligence Lab for Enterprises (SCALE@NTU), Nanyang Technological University, Singapore. His research interests include multi-agent systems, distributed systems, network security and algorithms.