Abstract:
Robots that are capable of offering high flexibility and high capability to solve the inherent system uncertainties, unknowns, and exceptions are increasingly demanded in...Show MoreMetadata
Abstract:
Robots that are capable of offering high flexibility and high capability to solve the inherent system uncertainties, unknowns, and exceptions are increasingly demanded in assembly tasks. It requires the robot to perceive, detect and locate its operating target. In this paper, we propose an object-localization method using natural features with a monocular camera. The object-localization includes three steps: initial localization using SURF detection and matching, probability-based natural right angle corner detection for pose estimation, and final adjustment using template matching. The proposed method deals with targets of different scales and rotations. It doesn't require extra sensors except a monocular camera, or extra artificial markers. Furthermore, the experimental results show the low average localization errors and high success rate.
Date of Conference: 03-07 December 2016
Date Added to IEEE Xplore: 02 March 2017
ISBN Information: