Scan Matching

Scan matching is the process of aligning pairs of point clouds, typically from lidar sensors, to estimate the relative pose (position and orientation) of a sensor between scans. Current research focuses on improving the speed and accuracy of scan matching algorithms, including those based on Normal Distributions Transform (NDT), branch-and-bound methods, and iterative closest ellipsoidal transforms (ICET), often leveraging GPU acceleration and incorporating additional sensor data like inertial measurements or visual features. These advancements are crucial for applications like autonomous navigation, robotics, and 3D mapping, enabling more robust and efficient localization and map building in diverse environments. Furthermore, research is addressing challenges like perspective errors and shadowing effects to enhance the reliability of scan matching in complex scenarios.