Deep Learning for Planetary Rover Localization

点击翻译按钮获取中文摘要

We are continuing to apply machine learning to problems in space research/exploration and establish deep learning computational hardware capabilities at NAOJ.In FY2022, we presented a paper "Physics Informed RNN-DCT Networks for Time-Dependent Partial Differential Equations" at the International Conference on Computational Science (ICCS). This work incorporates techniques from the signal processing domain into physics-informed neural networks, greatly improving accuracy and speed in fluid simulations.We also published a paper "Low-thrust rendezvous trajectory generation for multi-target active space debris removal using the RQ-Law" to Advances in Space Research. We use path optimization algorithms to find the best orbital trajectory for a spacecraft to remove space debris objects.