Real-time Lunar Landing Site Detection Pipeline Using YOLOv8 and Kalman Filtering

rsasaki0109 · x · 2026-08-23

A real-time computer vision pipeline that detects craters in a video feed using a custom-trained YOLOv8 model. It evaluates surrounding terrain to identify and track a safe, flat, well-lit landing site using vectorized slope/luminance analysis, a convex-hull safe-zone estimate, and Kalman filtering for smooth target tracking.

Related event: YOLOv8-based system autonomously selects lunar landing sites(2 posts)→

Original post →

More from Embodied

Embodied channel →