Overview

Multi-sensor fusion technique has made great success and been the underpinning for many applications and systems in the last decades. However, with the rapid development of sensor technologies and recent data explosion of information systems, multi-sensor fusion community has been facing big challenges in making full use of diverse sources of information, efficiently modelling high-dimensional data, reasoning uncertainty from imperfect data, etc.

Recent advances in deep learning have revolutionized some fields, e.g., computer vision, and show potential capabilities to benefit multi-sensor fusion, though it is not clear how deep learning can naturally model uncertain and explicitly use prior knowledge which are key for multi-sensor fusion. As sensor fusion and deep learning are becoming two main components in many robotic systems, industrial automation, cyber-physical systems, internet of things and so on, it is inevitable that they will interact with each other closer. However, as an emerging area, extensive novel research and practical applications are needed to explore the optimal integration of these two techniques and tackle the challenges.

The main objective of the workshop is to bring researchers of interest in these two areas to address key challenges in both domains, identifying limitations, ideally envisioning possible approaches to solve them, and opportunities, with the potential to advance both of techniques.

Invited Speakers

Dr. Xuetao Zhang
Dr. Xuetao Zhang

Associate Professor, Dalian University of Technology

Dr. Xuetao Zhang is currently an Associate Professor at the Intelligent robotic Laboratory, Dalian University of Technology, China. He received his Ph.D. in 2020 from Nankai University with the Institute of Robotics and Automatic Information System (IRAIS). His research interests include motion planning, visual servoing, state and disturbance estimation for rotorcraft Unmanned Aerial Vehicles.

Dr. Bo Yang
Dr. Bo Yang

Assistant Professor, University of Oxford and Hong Kong Polytechnic University

Dr. Bo Yang is an incoming Assistant Professor at The Hong Kong Polytechnic University which he will be joining in November 2020. He completed his DPhil degree from the Department of Computer Science at University of Oxford. His research interests lie in deep learning, computer vision, and robotics.

Programme

Time (UTC) Topic
14:00-14:05 Welcome and Introduction
14:05-14:50 Dr. Xuetao Zhang
Title: Visual Servoing of Rotorcraft Unmanned Aerial Vehicles
Abstract: Vision-based rotorcraft unmanned aerial vehicle (UAV) system has attracted great interest from both academia and industry, which has many applications including aerial photography, monitoring, rescue and so on. This talk mainly introduces the recent research results in the field of visual servoing for UAVs, which includes vision-based pose estimation, real-time motion planning under input constraints, and dynamic feedback control. On this basis, I will summarize the previous work and forecast the future trends of the rotorcraft UAVs. Finally, I will briefly introduce other related research results at Dalian University of Technology in terms of direct visual-inertial odometry, real-time moving target tracking, trail detection and scene understanding, and flying through a narrow gap, etc. And some preliminary results are demonstrated via videos.
14:50-15:35 Dr. Bo Yang
Title: Learning to Understand 3D Point Clouds
Abstract: Enabling machines to understand 3D scenes is a fundamental necessity for autonomous driving, augmented reality, and robotics. Core problems on 3D geometric data such as point clouds include semantic segmentation and instance segmentation. Due to the unique properties (i.e., unstructured, irregular, and orderless) of 3D point clouds, it is highly challenging to achieve fast and robust segmentation. In this talk, we will firstly present a brief review of existing neural algorithms for point cloud analysis. We will then provide the details of two novel neural architecture RandLA-Net and 3D-BoNet, particularly for large-scale point cloud semantic segmentation and instance segmentation.
15:35-16:05 Q&A and Discussion
16:05-16:10 Concluding Remarks

Organisers

Dr. Sen Wang , Associate Professor at Heriot-Watt University and Edinburgh Centre for Robotics, UK
Prof. Yan Zhuang , Professor at Dalian University of Technology, China