PaPEGAN: Parallelized Pose Estimation using Generative Adversarial Networks

Space and space-exploration have always provided mankind with a sense of mystery and aura. In the pursuit of quelling this curiosity man has invented spacecrafts to look into the depths of the cosmos. In this paper the researcher presents a novel architecture that combines the technologies of Generative Adversarial Networks (GANs), process-based parallelism and Convolutional Neural Network (CNN) based Pose Estimation of various spacecraft (for instance satellites and spaceships). The GANs with its generator and discriminator provides an increased number of images to augment the training possible using current datasets. The successful implementation of PaPEGAN will significantly advance the field of spacecraft pose estimation, in terms of accuracy and efficiency. The impact of this research extends to various space exploration applications, such as satellite navigation, robotic spacecraft control, autonomous landing, and trajectory planning, ultimately enhancing the success and effectiveness of space missions. Overall, the problem is to develop and demonstrate the effectiveness of PaPEGAN as an advanced solution for spacecraft pose estimation, revolutionizing the capabilities and performance of navigation and control systems in the realm of space exploration.