Perceptor

Perceptor Node

ros2 run kompass_ros perceptor

parameter name	type	default value	description
yolo_model	string	models/yolov8n-seg.pt	path to segmentatio model yolo v8 model weights
image_height	int	240	image height
image_width	int	240	image width
conf_thres	float	0.45	confidence threshold, between 0 and 1
iou_thres	float	0.45	confidence threshold, between 0 and 1
max_det	int	50	maximum detections allowed per image
tracker	string	botsort_params.yaml	configurations for yolo available tracker
moving_objects_only	bool	False	detects objects that can move
server_uri	string	http://localhost:8000	perception server URI
server_timeout	float	10.0	request timeout
debug	List of comma separated string	image	stuff that are possible to visualize/debug
detections_topic	string	detections	detections topic name
image_topic	string	intel_realsense_r200_depth/image_raw	RGB image topic name
depth_topic	string	intel_realsense_r200_depth/depth/image_raw	depth map topic name
camera_info_topic	string	intel_realsense_r200_depth/camera_info	camera info topic name
camera_frame	string	camera_rgb_optical_frame	camera tf frame
base_frame	string	base_link	robot tf frame

message type	default topic name	description
sensor_msgs/Image	image_raw	RGB image
sensor_msgs/Image	depth_raw	aligned depth map
[sensor_msgs/CameraInfo	camera_info	camera information (intrinsics K matrix)

message type	default topic name	description
kompass_interfaces/detections	/detections	detected semantics inside the current RGB frame and projected using the aligned depth
sensor_msgs/Image	detections_image	image with detections and bounding boxes