Pattern Detector

The pattern_detector package is designed for detecting and visualizing patterns in point clouds collected from a Lidar sensor. The package utilizes the Robot Operating System (ROS) for handling sensor data and visualization.

Overview

• Structures: This part of the package includes classes that define the primary data structures used by the detector, such as Point, Quaternion, Pose, PointCloud, Cluster, SubCluster, ValidCluster, and DetectedShape.

• PointCloudProcessor: This is a helper class that takes in the raw point cloud data and applies various processing steps. This includes clustering, subclustering, filtering, and finally detecting shapes within the point cloud.

• RVizVisualizer: This class is responsible for visualizing the detection results in RViz. It publishes markers for clusters, subclusters, valid clusters, and the detected shape.

• PatternDetectorROS: This is the primary class which orchestrates the entire detection process. It subscribes to the Lidar scan data, processes it, visualizes the results, and publishes the detected shape.

Configuration

The behavior of the pattern detector can be adjusted via a set of ROS parameters. Here's an overview:

• frame_id: Frame id to be used by the detector.
• scan_topic_name: Name of the topic from which scan data is subscribed.
• cluster, subcluster, valid_clusters: These sections contain parameters for adjusting the clustering and visualization process such as marker_topic_name, marker_color, eps, min_samples, and id.
• parking_pose: Topic name for publishing the parking pose.
• marker_timelife: Lifetime of the markers in RViz.
• max_point_distance: Maximum distance from the sensor for points to be included in the point cloud.
• scan_angle: The minimum and maximum angles to be included in the scan.
• wall_filter: Parameters for RANSAC wall filtering.
• simple_pattern: Parameters to define the pattern for detection, such as pattern_width, pattern_offset, and required_subclusters.
• distance_from_centroid: The maximum distance from the centroid for a point to be considered as part of a cluster.
• pose_buffer: Parameters for pose estimation, including max_samples, confirm_threshold, and distance_threshold.

Usage

Once you have adjusted the parameters to suit your setup, you can start the pattern detector with the following command:

roslaunch pattern_detector pattern_detector.launch

You can then visualize the results in RViz by subscribing to the relevant marker topics.

Note

Please adjust the parameters according to your sensor setup and environment for optimal results. This package is tested with ROS Noetic and Python 3.8.