Distance estimation with 2.5D anchors and its application to robot navigation

ROBOMECH Journal

article

Hirotaka Hachiya¹  Yuki Saito²  Kazuma Iteya¹  Masaya Nomura¹  Takayuki Nakamura¹ 

[1] Faculty of System Engineering, Wakayama University;Department of Statistical Science, Graduate University for Advanced Studies

关键词: Deep learning;    Monocular camera image;    Distance estimation;    Navigation;   

DOI  :  10.1186/s40648-018-0119-5

学科分类：社会科学、人文和艺术（综合）

来源: Springer

【 摘 要 】

Estimating the distance of a target object from a single image is a challenging task since a large variation in the object appearance makes the regression of the distance difficult. In this paper, to tackle such the challenge, we propose 2.5D anchors which provide the candidate of distances based on a perspective camera model. This candidate is expected to relax the difficulty of the regression model since only the residual from the candidate distance needs to be taken into account. We show the effectiveness of the regression with our proposed anchors, by comparing with ordinary regression methods and state-of-the-art 3D object detection methods, through Pascal 3D+ TV monitor and KITTI car experiments. In addition, we also show an example of practical uses of our proposed method in a real-time system, robot navigation, by integrating with ROS-based simultaneous localization and mapping.

【 授权许可】

CC BY