In the construction and maintenance for large space equipment, it is essential to ensure the control accuracy and improve the dexterity of the space manipulator. In this paper, a Finite-Time Convergence Kinematic Control (FTCKC) added with Acceleration Level Dexterity Optimization (ALDO) scheme is proposed to solve the kinematic uncertainty and dexterity optimization problems of redundant space manipulators. Concretely, distinguishing from the asymptotic convergence property of traditional adaptive Jacobian methods, the FTCKC scheme is adopted to construct the equality constraint to address the model uncertainty problem, and its error can converge within a finite time. Subsequently, the dexterity index is reconstructed at acceleration level by a multi-level target handling method. Then, the equality constraint, optimization task, and limit constraints are reformulated as a quadratic programming problem. Moreover, a Recurrent Neural Network (RNN) is engineered for the constructed FTCKC-ALDO scheme. Finally, the superiority of the FTCKC-ALDO-RNN scheme is verified by experiments.

This study focuses on addressing kinematic singularity analysis and avoidance issues for a space station remote manipulator system (SSRMS)-type reconfigurable space manipulator. The manipulator is equipped with a non-spherical wrist and two lockable passive telescopic links (LPTLs), which enable it to have both active revolute and passive prismatic joints and operate in two distinct modes. To begin with the kinematic singularity analysis, the study derives the differential kinematic equations for the manipulator and identifies the dominant Jacobian matrix that causes singularities. Subsequently, an in-depth analysis of singularities from multiple perspectives is conducted. Firstly, a kinematic singularity map method is proposed to capture the distribution of singularities within the reachable workspace. Then, the influence of the two LPTLs on singularities is thoroughly examined. Finally, a new method based on the matrix rank equivalence principle is introduced to determine singularity conditions, enabling the identification of all the singular configurations for the SSRMS-type reconfigurable manipulator. Notably, this method significantly reduces computational complexity, and the singularity conditions obtained have more concise equations. For the singularity avoidance problem, a novel method is developed, which simultaneously addresses the requirements of real-time performance, high precision, and the avoidance of both kinematic singularities and joint limit constraints. Benefiting from these excellent properties, the proposed method can effectively resolve the singularity issues encountered separately by the SSRMS-type reconfigurable manipulator in its two operational modes. Several typical simulations validate the utility of all the proposed methods.

Landing mechanism tends to rebound and turn over, and the stability time is long when landing on the small celestial body. Its landing performance in different conditions is necessary to be evaluated to guide the landing. Here, landing performance evaluation is realized by simulation. Key factors affecting the landing performance including cardan element damping, foot anchors, retro-rocket thrust, landing slope angle, and landing attitude are analyzed. A microgravity platform is built to test the landing mechanism, and the consistency between the simulation and the experiment is compared. On the basis of simulation and experiment, some landing suggestions are proposed to improve the landing performance.