Globally Stable Velocity Estimation Using Normalized Velocity Measurement
Journal article, Peer reviewed
Accepted version
Åpne
Permanent lenke
http://hdl.handle.net/11250/2636918Utgivelsesdato
2019Metadata
Vis full innførselSamlinger
Originalversjon
10.1177/0278364919887436Sammendrag
The problem of estimating velocity from a monocular camera and calibrated inertial measurement unit (IMU) measurements is revisited. For the presented setup, it is assumed that normalized velocity measurements are available from the camera. By applying results from nonlinear observer theory, we present velocity estimators with proven global stability under defined conditions, and without the need to observe features from several camera frames. Several nonlinear methods are compared with each other, also against an extended Kalman filter (EKF), where the robustness of the nonlinear methods compared with the EKF are demonstrated in simulations and experiments.