Skip to content

low_intensity_cluster_filter#

Purpose#

The low_intensity_cluster_filter is a node that filters clusters based on the intensity of their pointcloud.

Mainly this focuses on filtering out unknown objects with very low intensity pointcloud, such as fail detection of unknown objects caused by raindrop or water splash from ego or other fast moving vehicles.

Inner-workings / Algorithms#

Inputs / Outputs#

Input#

Name Type Description
input/object tier4_perception_msgs::msg::DetectedObjectsWithFeature input detected objects

Output#

Name Type Description
output/object tier4_perception_msgs::msg::DetectedObjectsWithFeature filtered detected objects

Parameters#

Name Type Description Default Range
intensity_threshold float The threshold of average intensity for filter. 1.0 N/A
existence_probability_threshold float The existence probability threshold to apply the filter. 0.2 N/A
max_x float Maximum of x of the filter effective range. 60.0 N/A
min_x float Minimum of x of the filter effective range. -20.0 N/A
max_y float Maximum of y of the filter effective range. 20.0 N/A
min_y float Minimum of y of the filter effective range. -20.0 N/A
filter_target_label.UNKNOWN boolean If true, unknown objects are filtered. true N/A
filter_target_label.CAR boolean If true, car objects are filtered. false N/A
filter_target_label.TRUCK boolean If true, truck objects are filtered. false N/A
filter_target_label.BUS boolean If true, bus objects are filtered. false N/A
filter_target_label.TRAILER boolean If true, trailer objects are filtered. false N/A
filter_target_label.MOTORCYCLE boolean If true, motorcycle objects are filtered. false N/A
filter_target_label.BICYCLE boolean If true, bicycle objects are filtered. false N/A
filter_target_label.PEDESTRIAN boolean If true, pedestrian objects are filtered. false N/A

Assumptions / Known limits#

(Optional) Error detection and handling#

(Optional) Performance characterization#

(Optional) Future extensions / Unimplemented parts#