Comparing holistic and feature-based visual methods for estimating the relative pose of mobile robots

详细信息    查看全文

• 作者：David Fleer ; ^{dfleer@ti.uni-bielefeld.de}Author Vitae ; Ralf Mö ; ller ^{moeller@ti.uni-bielefeld.de}Author Vitae
• 关键词：Visual pose estimation ; Robot localization ; Local visual features ; Holistic visual homing ; Image databases
• 刊名：Robotics and Autonomous Systems
• 出版年：2017
• 出版时间：March 2017
• 年：2017
• 卷：89
• 期：Complete
• 页码：51-74
• 全文大小：2442 K
• 卷排序：89

文摘

Local features and holistic methods for visual robot-pose estimation are compared. Holistic results are good, some feature methods are more precise but also slower. Pose estimation accuracy strongly depends on distance between image locations. Under day–night illumination changes, the methods’ performances differ greatly. Computation times are provided for desktop and embedded robot platforms.