Monocular SLAM with Undelayed Initialization for Indoor Robot
Journal
Robotics and Autonomous Systems
Date
2012.06.01
Abstract
This paper presents a new feature initialization method for
monocular EKF SLAM (Extended Kalman Filter Simultaneous
Localization and Mapping) which utilizes a 3D measurement
model in the camera frame rather than 2D pixel coordinates in
the image plane. The key idea is to regard a camera as a
range and bearing sensor, of which range information have a
lot of uncertainties. 2D pixel coordinates of measurement are
converted to 3D points in the camera frame with assumed depth
and the element of the measurement noise covariance
corresponding to the depth of the feature is set a very large
value. The proposed method provides immediate feature
initialization without any delay and runs within the EKF
framework without the need of extra procedures to estimate
the feature depth. Furthermore, it has very low linearization
errors, which can be critical for the performance of EKF.
Experiments show robust operation in indoor environments with
computational efficiency.