Real-time 6DOF (6 Degree of Freedom) pose estimation of an uncooperative spacecraft is an important part of proximity operations, e.g., space debris removal, spacecraft rendezvous and docking, on-orbit servicing, etc. In this paper, a novel efficient deep learning based approach is proposed to estimate the 6DOF pose of uncooperative spacecraft using monocular-vision measurement. Firstly, we introduce a new lightweight YOLO-liked CNN to detect spacecraft and predict 2D locations of the projected keypoints of a prior reconstructed 3D model in real-time. Then, we design two novel models for predicting the bounding box (bbox) reliability scores and the probability of keypoints existence. The two models not only significantly reduce the false positive, but also speed up convergence. Finally, the 6DOF pose is estimated and refined using Perspective-n-Point and geometric optimizer. Results demonstrate that the proposed approach achieves 73.2% average precision and 77.6% average recall for spacecraft detection on the SPEED dataset after only 200 training epochs. For the pose estimation task, the mean rotational error is 0.6812 • , and the mean translation error is 0.0320m. The proposed approach achieves competitive pose estimation performance and extreme lightweight (∼ 0.89 million learnable weights in total) on the SPEED dataset while being efficient for real-time applications.

A novel satellite target recognition method based on radar data partition and deep learning techniques is proposed in this paper. For the radar satellite recognition task, orbital altitude is introduced as a distinct and accessible feature to divide radar data. On this basis, we design a new distance metric for HRRPs called normalized angular distance divided by correlation coefficient (NADDCC), and a hierarchical clustering method based on this distance metric is applied to segment the radar observation angular domain. Using the above technology, the radar data partition is completed and multiple HRRP data clusters are obtained. To further mine the essential features in HRRPs, a GRU-SVM model is designed and firstly applied for radar HRRP target recognition. It consists of a multi-layer GRU neural network as a deep feature extractor and linear SVM as a classifier. By training, GRU neural network successfully extracts effective and highly distinguishable features of HRRPs, and feature visualization technology shows its advantages. Furthermore, the performance testing and comparison experiments also demonstrate that GRU neural network possesses better comprehensive performance for HRRP target recognition than LSTM neural network and conventional RNN, and the recognition performance of our method is almost better than that of other several common feature extraction methods or no data partition.

Optical data are important data sources for operating state estimation of space targets. Inversion of the rotation rate of tumbling targets via light curves can help us perform on-orbit services. However, previous methods cannot estimate the precession and spin rates of tumbling targets from light curves simultaneously. To solve the problem, an efficient precession and spin rate estimation algorithm for tumbling targets is proposed in this paper. The method combines the variational mode decomposition (VMD) method and mutual information (MI). Specifically, VMD is utilized to decompose each light curve into discrete frequency intrinsic mode functions (IMFs). Then, the MI between each IMF and light curve is generated, and the IMF frequencies corresponding to the two maximum MI values are extracted as spin and precession frequencies. Finally, the two frequencies are converted into spin and precession rates. Experimental results show that the estimation accuracy of the precession and spin rates is no less than 97% for a small nutation angle (no greater than 20º). The method provides a simple way to invert a space target state and mine more information from light curves.

No abstract

With the development of Space Domain Awareness(SDA), satellites’ optical characteristics are becoming attention-grabbing. Sunlight was usually considered the only light source for the satellites. However, in the actual observation, researchers have found that earthshine and moonlight would increase errors of the observation results, which have greatly influence the estimation of the satellite’s state. In order to avoid this influence, we propose an observation strategy. Firstly, we propose an accurate earthshine model, which considers the earth’s volume and favors long-time continuous satellite observation. Then, we explore the earthshine and moonlight’s impact on satellite observation results and find that this impact varies with the satellite attributes. Furthermore, we Figure out the law of this impact and establish a connection between this law and observation geometry. Finally, a Period Contribution model is proposed to provide a corresponding observation strategy to avoid the influence of earthshine and moonlight.