To address challenges in takeover control spacecraft systems involving a large number of cellsats, module heterogeneity, local communication constraints, communication delays, and uncertainties and external disturbances, a fully distributed control method with integrated control and allocation is proposed. First, the attitude tracking error kinematics and dynamics model of the combined spacecraft is transformed into a State Dependent Coefficients (SDC) model. Then, based on the system model with an augmented input matrix, the traditional Tube Model Predictive Control (TMPC) framework is modified according to the characteristics of the control allocation problem. By combining the Delay-Tolerant Augmented Consensus Tracking Alternating Direction Method of Multipliers (DTAC-ADMM) distributed optimization algorithm under communication delays, the cooperative attitude control problem is transformed into a multi-decision-variable optimization problem with coupled constraints. The proposed framework explicitly considers actuator constraints, dynamic constraints of the attitude tracking error under prescribed performance conditions, as well as communication delays and packet losses among cellsats modules. Consequently, each module can independently compute its required control torque using only information from its neighboring modules. Finally, simulations are conducted to verify the correctness and effectiveness of the proposed control scheme, demonstrating its suitability for co-operative control in heterogeneous satellite clusters with only local communication capabilities. Compared with the traditional two-layer “control + allocation” framework, the proposed single-layer framework eliminates the need for a centralized controller for computation and allocation, and can accommodate the online addition or removal of cellsats during control. It fully exploits the control capability of each module, avoids actuator saturation, resolves the strong coupling difficulties of traditional TMPC in control allocation, achieves global optimization of total control torque energy consumption, and exhibits strong disturbance rejection capability. Moreover, the distributed optimization remains applicable even in the presence of communication delays and packet losses. The proposed method achieves fully distributed, simplifies controller parameter tuning, provides strong robustness, and is suitable for practical engineering applications.

Non-cooperative spacecraft operating in high-value orbits pose significant threats to space security, as ground-based navigation and orbit determination systems cannot effectively acquire key information such as payloads and attitudes, making it difficult to assess their functionality and threat level. Traditional passive observation methods suffer from security concerns and long observation cycles. To address these limitations, this paper proposes an active observation trajectory planning method that integrates a state-gain reachable set-based collision avoidance strategy with Fast Model Predictive Control (FMPC). To tackle the issue of rapidly declining safety in traditional navigation error ellipsoids over time, the proposed method utilizes an analytical approximation of the geometric bounds of the spacecraft's size-expanded state-gain reachable. This enables the development of a collision avoidance strategy that balances high computational efficiency with enhanced safety. Additionally, to overcome the challenges of prolonged observation cycles and insufficient mission concealment in traditional free-flight approaches, a reference trajectory is designed to ensure the sensor's field of view fully covers the target body and its critical payload. FMPC is then employed to calculate trajectory tracking control laws in real time, satisfying the constraints of mission cycles and forming a trajectory planning strategy that combines short cycles with comprehensive information acquisition. Compared to traditional methods, this approach achieves significant improvements in safety, concealment, and observation efficiency, providing an effective solution for the identification of non-cooperative on-orbit targets.

Aiming at the interception problem of noncooperative evader spacecraft adopting random maneuver strategy in one-to-one orbital pursuit–evasion problem, an interception strategy with decision-making training mechanism for the pursuer based on deep reinforcement learning is proposed. Its core purpose is to improve the success rate of interception in the environment with high uncertainty. First of all, a multi-impulse orbit transfer model of pursuer and evader is established, and a modular deep reinforcement learning training method is built. Second, an effective reward mechanism is proposed to train the pursuer to choose the impulse direction and impulse interval of the orbit transfer and to learn the successful interception strategy with the optimal fuel and time. Finally, with the evader taking a random maneuver decision in each episode of training, the trained decision-making strategy is applied to the pursuer, the corresponding interception success rate of which is further analyzed. The results show that the pursuer trained can obtain universal and variable interception strategy. In each round of pursuit–evasion, with random maneuver strategy of the evader, the pursuer can adopt similar optimal decisions to deal with high-dimensional environments and thoroughly random state space, maintaining high interception success rate.