Multi-Data Fusion and intelligent connectivity enhancing autonomous transport for users

多数据融合是一项重要的软件支持技术，用于增强日常生活中广泛关键应用中的实时传感器和机器人平台。该项目将研究图形处理单元（GPU）上立体相机和激光雷达数据的嵌入式融合，以提供一种智能和智能的方法，通过使用鲁棒估计来实时获取3D信息，清除伪影。这将在宽视场（FoV）上创造一种新型的实时机器人3D视觉传感器。立体成像可以覆盖宽视场，并提供从微观到轨道距离的密集3D点云数据，但存在计算强度高和距离分辨率有限的问题。通常，激光雷达在3D生成方面是精确和快速的，但它的缺点是视场狭窄，信息密度低，在远距离上效率低下。

Multi-data fusion is an essential software-enabling technology for enhancing real-time sensors and robotic platforms in a wide range of critical applications for use in everyday life. This project will investigate the embedded fusion of stereo-camera and LIDAR data on Graphics Processing Units (GPU) to provide a smart and intelligent means of obtaining 3D information in real-time, cleansed of artefacts through the use of robust estimation. This will create a new class of real-time robotic 3D vision sensors over a wide Field of View (FoV). Stereo imaging can cover a wide FoV and provide dense 3D point cloud data from microscopic to orbital distances, but suffers from high computational intensity and limited range resolution. LIDAR, typically, is precise and fast for 3D generation, but suffers from a narrow FoV, less dense information and is inefficient at large distances.