Skip to main content

LiDAR for Humanoids

What is LiDAR?​

Light Detection and Ranging (LiDAR) uses laser pulses to measure distances to objects. For humanoids, LiDAR is crucial for:

  • SLAM (Simultaneous Localization and Mapping): Building a map of the environment.
  • Obstacle Avoidance: Detecting walls, people, and furniture.

Types of LiDAR​

  1. 2D LiDAR: Scans a single plane (e.g., RPLIDAR). Good for flat floors.
  2. 3D LiDAR: Scans a volume (e.g., Velodyne, Ouster). Essential for humanoids to see overhangs and complex terrain.

Integrating LiDAR with ROS 2​

We typically use the sensor_msgs/LaserScan (2D) or sensor_msgs/PointCloud2 (3D) message types.

# Example: Launching a Velodyne driver
ros2 launch velodyne_driver velodyne_driver_node-VLP16-launch.py

Processing Point Clouds​

We use PCL (Point Cloud Library) to filter and segment data.

// C++ snippet to downsample a point cloud
pcl::VoxelGrid<pcl::PointXYZ> sor;
sor.setInputCloud(cloud);
sor.setLeafSize(0.01f, 0.01f, 0.01f);
sor.filter(*cloud_filtered);