当前位置: 云顶集团手机版 > 学术报告 > 正文

从几何学到深度学习的视觉同步定位与地图构建

发布时间:2020-07-24 09:27:58 发布人:孟斌  审核人:李天镇

报告题目:从几何学到深度学习的视觉同步定位与地图构建

报告时间:2020年7月30日下午4点

报告形式:Zoom在线讲座

Zoom ID: 94577065664

主讲人:人工智能、深度学习领域知名专家

主办方:综合交通运输协同创新中心

报告摘要:

视觉同步定位与地图构建(Visual SLAM)是一种使用平台搭载相机的测量结果,来估计移动平台位置和方向时间变化的技术。该技术吸引了大量研究者关注,正逐渐在各种潜在应用中得到推广。然而,当把精度、实时性、鲁棒性等性能因素纳入考虑之中时,在工程上实现视觉同步定位与地图构建技术却面临诸多挑战。本报告首先从几何方法角度、然后从深度学习角度讲述视觉建图与定位技术的国内外发展现状,其中介绍的基于深度学习的方法覆盖了监督视觉测程法、监督语义建图法、无监督视觉同步定位与地图构建法等在内的不同方法。最后,报告人还将指明在这一工作领域的未来研究方向。

Visual SLAM is a technique to estimate the change of a mobile platform in position and orientation over time by using the measurements from on-board cameras. It attracts significant attentions from large number of researchers and is gaining the popularity in various potential applications. However, it is very challenging in both of technical development and engineering implementation when accuracy, real time performance, robustness, and operation scale are taken into consideration. This talk is to report the state of the art visual mapping and localisation techniques starting from the perspective of Geometry approaches and them moving towards Deep Learning approaches. Our work on applying deep neural networks to visual mapping and localisation research will be presented. Some recent results will be demonstrated during the talk.

XML 地图 | Sitemap 地图