• 

  钱辉环

  优德官网副院长、香港中文大学（深圳）助理教授

  执行主席

  Bio

  Abstract
• 

  肖文力

  香港中文大学（深圳）本科生

  Tackling Safe and Efficient Multi-Agent Reinforcement Learning via Dynamic Shielding

  Bio

  Wenli Xiao is a senior majoring in Electrical Information Engineering (Computer Engineering stream) at the Chinese University of Hong Kong, Shenzhen. He has interned at the Robotics Institute Summer Scholar (RISS) at Carnegie Mellon University, where he collaborated with Prof. John Dolan and Yiwei Lyu on safe Multi-Agent Reinforcement Learning. He was also a research assistant at the NCEL Lab at the Shenzhen Institute of Artificial Intelligence and Robotics for Society (优德官网), where he did research in Federated Learning for Robotics and IoT, mentored by Prof. Jianwei Huang and Prof. Bing Luo. His current research interests lie in Autonomous Systems, Robotics, and Reinforcement Learning.

  Abstract

  Multi-Agent Reinforcement Learning (MARL) has been increasingly used in safety-critical applications but has no safety guarantees, especially during training. In this paper, we propose dynamic shielding, a novel decentralized MARL framework to ensure safety in both training and deployment phases. Our framework leverages Shield, a reactive system running in parallel with the reinforcement learning algorithm to monitor and correct agents’ behavior. In our algorithm, shields dynamically split and merge according to the environment state in order to maintain decentralization and avoid conservative behaviors while enjoying formal safety guarantees. We demonstrate the effectiveness of MARL with dynamic shielding in the mobile navigation scenario.

• 

  林沐晗

  香港中文大学（深圳）本科生

  Less is more: A Robust Visual Inertial Odometry with Active Feature Extraction

  Bio

  Muhan Lin is currently an undergraduate student at the Chinese University of Hong Kong (Shenzhen), majoring in Computer Engineering. Her research interest is in computer vision, localization, sensor data fusion, multi-robot cooperation, and path planning, which are the basis of making robots realize complex but robust group tasks. As a rising researcher in robotics, Muhan cooperated with Dr. Yue Wang on robot-to-robot relative localization at Shenzhen Institute of Artificial Intelligence and Robotics for Society, advised by Prof. Tim Lun Lam. She  then extends her research to Visual Odometry and works on this with Shibo Zhao at the AirLab at Carnegie Mellon University, advised by Prof. Sebastian Scherer. She explores pushing the boundary of Visual Inertial Odometry in challenging environments utilizing her experience in relative pose estimation.

  Abstract

  To achieve robust performance, it is common for visual odometry and SLAM to track more features, like several hundreds of points in real time. Although this strategy performs well on high-end desktop PCs, it is difficult to apply it to some mobile platforms with limited computation resources, such as VR, Micro UAV, and multi-camera systems. Additionally, noisy visual feature points may decrease the accuracy of visual odometry and SLAM. Therefore, fewer but more informative features can boost efficiency and accuracy compared to extracting more features. It means that less is more. To achieve this target, we propose a new criterion for the active feature selection based on singular values and then incorporate this method into an advanced VIO system, TP-TIO [1]. With the new system, using half of the features required by the original TP-TIO, the residuals can be reduced to 56.23% of the ones generated by the original TP-TIO without increasing the processing time to a large degree. The new system was verified with the mmpug datasets [2], which were extracted in a long and dark corridor.

视频回首