单目相机-轮式里程计的自动内外参标定算法

本文提出了一种新的单目相机和轮式里程计系统的内、外参标定方法。不同于目前单独标定内外参的方法,本文方法能在一个平台上同时标定内参和外参。平台包含4个棋盘格,其中3个棋盘格基于优化理论标定内参,第4个棋盘用于标定外参。基于该平台,相机-里程计系统只需对3个棋盘格拍摄一次进行内参标定,然后沿直线行走,连续拍摄第4个棋盘格完成外参初始标定,最后根据最小二乘优化外参。通过使用扫地机器人采集的真实数据集进行测试,证明了该方法能精确完成内参的标定并自动评估标定质量,具有有效性和优越性。

Automatic intrinsic and extrinsic calibration for a monocular camera-odometry system

In this paper, we provide a novel intrinsic and extrinsic calibration method for a monocular camera and odometry system automatically. Unlike previous methods aiming at only alternative intrinsic or extrinsic parameters, this method efficiently calibrates both intrinsic and extrinsic parameters using one platform in a unified way. Our platform contains four chessboards with three of them assembled in an optimized way for intrinsic calibration, and the fourth chessboard for the extrinsic calibration. Based on this platform, the camera-odometry system only need to take one shot image of the three chessboards for intrinsic calibration, and then move with a planar z-path seeing the fourth chessboard for the extrinsic calibration. Testing on the real dataset collected from clear robot, we show that our method can accurately estimate the intrinsic parameters with a measurement to evaluate the calibration quality automatically, and it’s effficient and outperform the previous methods.