A ROS package with an implemented kalman filter, using the OpenCV KalmanFilter class.
The main node accepts geometry_msgs/PointStamped
points and it applies a Kalman filter. The input topic is defined in the config file params.yaml
. The filtered points are published to the topic /kalman_points
.
keypoint_topic
: Input topicvariable_timestamp
: True if the prediction is based on the timestamp of the messages (default: False)frequency
: Used in the prediction only ifvariable_timestamp
is set to False. (default: 30 => visual sensor's frequency)Rx, Ry, Rz
: Noise covariance matrixQ
: Process covariance matrix
- Run
roslaunch kalman_filter kalman.launch