Actively tracking an arbitrary space non-cooperative object relied on visual sensor remains a challenging problem. In this paper, we provide an open-source benchmark for space non-cooperative object visual tracking including simulated environment, evaluation toolkit, and a PBVS baseline algorithm, which can facilitate the research in this topic especially for those methods based on deep reinforcement learning. We also present an end-to-end active visual tracker based on deep Q-learning, named as DRLAVT, which learns approximately optimal policy merely took color or RGBD images as input. To the best of our knowledge, it is the first intelligent agent used for active visual tracking in aerospace domain. The experiment results show that our DRLAVT achieves excellent robustness and real-time performance compared with the PBVS baseline, benefitted from the design of complex neural network and efficient reward function. In addition, the multiple targets training adopted in this paper effectively guarantees the transferability of DRLAVT by forcing the agent to learn optimal control policy with respect to motion patterns of the target.

The capability of UAVs for efficient autonomous navigation and obstacle avoidance in complex and unknown environments is critical for applications in agricultural irrigation, disaster relief and logistics. In this paper, we propose the DPRL (Distributed Privileged Reinforcement Learning) navigation algorithm, an end-to-end policy designed to address the challenge of high-speed autonomous UAV navigation under partially observable environmental conditions. Our approach combines deep reinforcement learning with privileged learning to overcome the impact of observation data corruption caused by partial observability. We leverage an asymmetric Actor–Critic architecture to provide the agent with privileged information during training, which enhances the model’s perceptual capabilities. Additionally, we present a multi-agent exploration strategy across diverse environments to accelerate experience collection, which in turn expedites model convergence. We conducted extensive simulations across various scenarios, benchmarking our DPRL algorithm against state-of-the-art navigation algorithms. The results consistently demonstrate the superior performance of our algorithm in terms of flight efficiency, robustness and overall success rate.

Active tracking of space noncooperative object that merely relies on vision camera is greatly significant for autonomous rendezvous and debris removal. Considering its Partial Observable Markov Decision Process (POMDP) property, this letter proposes a novel deep recurrent neural network architecture, named as recurrent and attention module based active visual tracking (RAMAVT), incorporating Multi-Head Attention (MHA) module and Squeeze-and-Excitation (SE) layer that remarkably improve the representative ability of neural network with almost no extra computational cost. It has been successfully applied to value-based and policy gradient-based deep reinforcement learning algorithm, and learned to drive the chasing spacecraft to follow arbitrary space noncooperative object with high-frequency and near-optimal velocity control commands. Extensive experiments and robustness evaluations implemented on space non-cooperative object active tracking (SNCOAT) benchmark show the betterment and robustness of our method compared with other state-of-the-art active visual trackers. In addition, we make further ablation study and interpretability research on RAMAVT of which validity and rationality have been demonstrated.

Due to the adverse illumination in space, non-cooperative object perception based on multi-source image fusion is crucial for on-orbit maintenance and orbital debris removal. In this paper, we first propose a dual-branch multi-scale feature extraction encoder combining transformer block and Inception block to extract global and local features of visible and infrared images and establish high-dimensional semantic connections. Secondly, different from the traditional artificial design Fusion strategy, we propose a Feature Fusion Module called Cross-Convolution Feature Fusion (CCFF) Module, which can achieve image feature level fusion. Based on the above, we propose a Dual-branch Fusion Network based on Transformer and Inception (DFTI) for space non-cooperative object, which is an image fusion framework based on auto-encoder architecture and unsupervised learning. The fusion image can simultaneously retain the color texture details and contour energy information of space non-cooperative objects. Finally, we construct a fusion dataset of infrared and visible images for space non-cooperative objects (FIV-SNO) and compare DFTI with seven state-of-the-art methods. In addition, object tracking as a follow-up high-level visual task proves the effectiveness of our method. The experimental results demonstrate that compared with other advanced methods, the Entropy (EN) and average gradient (AG) of the fusing images using DFTI network are increased by 0.11 and 0.06 respectively. Our method exhibits excellent performance in both quantitative measures and qualitative evaluation.