Adaptive Edge Features Estimation for Humanoid Robot Visual Perception

摘要

Lidar scanner is a kind of sensor widely used in robotics visual perception, which provides accurate range data. Scan line grouping is an extremely fast plane segmentation and edge estimation method tailored for 2D Lidar scanners, which obtain 3D environment models by assembling 2D scan lines. In this paper, we propose an adaptive scan line split algorithm to overcome density shift problem of large scale scenes so that the scan line grouping method output more accurate plane segments and edges. The simulated experimental results indicate the proposed method is robust and promising in humanoid robotics applications.