Churinet - Applying Deep Learning for Minor Bodies Optical Navigation Articles uri icon

publication date

  • August 2023

issue

  • 4

volume

  • 59

International Standard Serial Number (ISSN)

  • 0018-9251

Electronic International Standard Serial Number (EISSN)

  • 1557-9603

abstract

  • This article presents Churinet, a hybrid neural network-based method, devoted to on-board spacecraft relative position and attitude estimation in the vicinity of minor bodies such as asteroids, comets or small moons, using monocular vision. In the context of navigating such minor bodies, traditional heuristic methods for spacecraft position and attitude determination encounter limitations in robustness and precision in the presence of adverse illumination conditions. Moreover, its performance is limited due to the computational cost resulting from the evaluation of a large number of possible pose hypotheses. In comparison, Churinet solves the relative pose estimation problem by directly learning the nonlinear transformation from a 2-D grayscale image to the 6-D pose vector space. Churinet is confirmed by a set of sequential convolutional neural networks (CNNs) organized in two levels. The high-level multiclass-classification CNN is in charge of determining the sector of the discretized 3-D space. Then, based on the sector estimation, the image is ingested by a low-level regression CNN, trained specifically for that sector, which estimates the pose of the camera. The secondary contribution of this research is the development of SPyRender, a tool for the generation of large sets of synthetic images, suitable for the training and testing of the designed CNNs. SPyRender implements GPU-accelerated physically based rendering, enabling the efficient generation of photorealistic images. SPyRender has been used with the 3-D model of comet 67P/C-G for producing multiple image sets covering the whole range of camera position, attitude, and illumination conditions, allowing us to study the impact of different geometries and image effects in the network performance.