Accurate measurement of panel and frame poses is essential for achieving automated and high-precision satellite assembly. Feature holes distributed across the satellite's mating surfaces typically serve as assembly and measurement reference points. However, the various types of feature holes, coupled with issues such as high reflectivity, edge occlusion, and surface defects, lead to the limitation of pose measurement accuracy and robustness based on feature holes. To address this challenge, a high-precision component pose measurement method based on stereo vision is proposed. First, a contour splitting and merging method based on geometric structural consistency (GSC) is developed to ensure low noise and high integrity of the contour. Next, to enhance feature-hole recognition accuracy and efficiency, a partitioned weighted sampling and iterative consistency (PWS-IC) strategy is introduced. Finally, to further improve measurement robustness, a component pose estimation method is proposed that effectively reduces the impact of outliers on pose estimation accuracy. Experimental results demonstrate that the measurement accuracy of feature holes reaches 0.03 mm, while the position and attitude estimation accuracy of satellite components attain 0.05 mm and $1.1<^>{\circ } \times 10<^>{-{5}}$ , respectively. Compared with existing methods, the proposed approach achieves substantial improvements in both accuracy and robustness. When applied to satellite assembly, the achieved mating accuracy reaches 0.15 mm, satisfying the technical requirement of 0.2 mm.

Semantic segmentation of 3D LiDAR point clouds is a crucial task in fields such as autonomous driving. LiDAR semantic segmentation based on range images has the advantage of being efficient. Nevertheless, when point clouds are projected onto range images, multiple points may be mapped onto a same grid which leads to boundary blurring and shape distortion. Projection also results in the low resolution in the vertical direction. In this study, we design a dual-branch encoder to address these challenges. The boundary enhancement branch uses the Scharr operator to counteract the blurring of edges that occurs when multiple points are projected onto a same grid. The backbone encoding branch employs multi-scale asymmetric residual blocks based on the Daubechies wavelet decomposition to minimize height information loss during downsampling for point clouds to range maps. Theses Daubechies asymmetric residual blocks also effectively expand the receptive field, and enhances the network's capability to extract features from range maps. Additionally, we introduce a bottom-up decoder that progressively fuses features of different resolutions through layered attention-guided fusion, enabling adaptive integration of high-level semantics with low-level boundary details. Experiments on the SemanticKITTI and SemanicPOSS demonstrate that our network attains mean Intersection over Union (mIoU) of 64.8 % and 52.9% respectively, outperforming most existing range-based methods.

Digital twin (DT) technology is changing the current pattern of intelligent manufacturing, it makes up for the shortcomings of process parameter optimization methods to improve real-time and predictability. This paper developed DT models for the robotic gluing system to predict the quality (width and thickness) of glue lines and optimize gluing parameters (trajectory and extrusion speeds). The DT framework based on the geometric, physical, behavioral, and rule models is constructed to monitor and optimize the gluing parameters in real-time. An improved backpropagation neural network (BPNN) prediction model based on whale optimization algorithm (WOA) is established to predict the width and thickness of glue lines from historical and real-time data, while simultaneously enabling real-time calculation of the cross-sectional area of glue lines. A multi-objective optimization model constructed using non-dominated sorting genetic algorithm (NSGA-II) is used to optimize the gluing parameters. The DT prototype of the robotic gluing system has been developed and verified experimentally. The position calibration of the geometric model is used to correct the gluing trajectory before gluing, and the position errors of the gluing points are within +/- 0.5 mm. The gluing trajectory is designed to test the effectiveness of the adaptive optimization of gluing parameters. The prediction errors of the width and thickness of the glue line are controlled between +/- 0.5 mm and +/- 0.3 mm, individually. After parameter optimization, the width and thickness of the glue line at the corner are reduced by 4.53 % and 7.54 %, respectively, thus avoiding glue accumulation. This reduction solves the problem of poor consistency in the quality of glue lines and verifies the feasibility of integrated monitoring, prediction, and optimization based on the DT model.

Fast readout unit (FRU) has been designed for generic control and data acquisition (DAQ) in heavy-ion nuclear experiments at the Heavy Ion Research Facility in Lanzhou (HIRFL) to reduce the development time, production cost, and maintenance difficulties of the readout electronics. Microtelecommunications computing architecture (MicroTCA) and MicroTCA systems are compatible with the FRU DAQ advanced mezzanine card (AMC) module. The FRU can connect four front-end readout electronics (FEE) for data collection, packaging, and transmission via optical links. The system’s backplane bus is a high-speed serial PCI Express (PCIe) bus, which is significantly faster than the conventional parallel bus. This article conducts the DAQ system’s performance test and application test. Thus, it has been demonstrated that the system can receive data over an optical link, transmit data over a PCIe link, and restore data. In addition, the AMC standard necessitates the implementation of the module management controller (MMC) onboard to monitor available and system-required hardware management parameters. Thus, the system can accommodate heavy-ion nuclear experiments. This article examines the FRU’s design and performance.

Bolt assembly by robots is a vital and difficult task for replacing astronauts in extra-vehicular activities (EVA), but the trajectory efficiency still needs to be improved during the wrench insertion into hex hole of bolt. In this paper, a policy iteration method based on reinforcement learning (RL) is proposed, by which the problem of trajectory efficiency improvement is constructed as an issue of RL-based objective optimization. Firstly, the projection relation between raw data and state-action space is established, and then a policy iteration initialization method is designed based on the projection to provide the initialization policy for iteration. Policy iteration based on the protective policy is applied to continuously evaluating and optimizing the action-value function of all state-action pairs till the convergence is obtained. To verify the feasibility and effectiveness of the proposed method, a noncontact demonstration experiment with human supervision is performed. Experimental results show that the initialization policy and the generated policy can be obtained by the policy iteration method in a limited number of demonstrations. A comparison between the experiments with two different assembly tolerances shows that the convergent generated policy possesses higher trajectory efficiency than the conservative one. In addition, this method can ensure safety during the training process and improve utilization efficiency of demonstration data.

The widely used kinesthetic demonstration method of dragging robotic manipulators cannot obtain reliable information for autonomous robot manipulation because an additional external force rather than pure contact force will be reflected on the force sensor in end-constrained manipulation tasks. Therefore, a noncontact robot demonstration method with human supervision is proposed to avoid external influence. A human demonstrator sends motion commands by mouse and observes force data reflected in a monitor to protect the robotic manipulator. Simultaneously, the human demonstrator supervises the position-orientation relationship between the end-effector and the manipulated object. The wrench insertion task is adopted to illustrate the advantage of the proposed method. A contact model is established according to demonstration data acquired from the proposed demonstration method, and an orientation adjustment strategy is verified. The strategy verification experiment illustrates not only the effectiveness of the simplified contact model and corresponding strategy but also the advantage of the proposed demonstration method.

Bolt screwing assembly task is a crucial part for robot fine manipulation. For widely applied flexible joints robot with six-dimensional force/torque sensor, data collected could not be seen as the contact force since it consists of internal force caused by deformation of the manipulator joints. Force analysis and geometric analysis could not be applied to judge the relative pose between robot end-effector and the environment. To this end, logistic regression method was used to classify the contact state by force signals. Besides, the criterion that bolt has entered the thread hole is also significant, while a boundary condition was proposed to solve this problem. A bolt screwing experiment was conducted to evaluate the proposed strategy, and the results demonstrate the effectiveness of the contact state classification and the boundary condition for judging the bolt entering the thread hole.

The contact force/torque between the end-effector of the space manipulator and the target spacecraft will reduce the efficiency and safety of the capture task. A capture strategy using PD-impedance combined control algorithm is proposed to achieve compliant contact between the chaser and target spacecraft. In order to absorb the impact energy, a spring-damper system is designed at the end-effector, and the corresponding dynamics model is established by Lagrange’s equation. Then a PD-impedance control algorithm based on steady-state force tracking error is proposed. Using this method, a compliant contact between the chaser and target spacecraft is realized while considering the dynamic coupling of the system. Finally, the general equation of the reference trajectory of the manipulator end-effector is derived according to the relative velocity and impact direction. The performance of the proposed capture strategy is studied by a co-simulation of MSC Adams and MATLAB Simulink in this paper. The results show that the contact plane at the end-effector of the manipulator can decelerate and detumble the target spacecraft. Besides, the contact force, relative velocity, and angular velocity all decrease to zero gradually, and the final stable state can be maintained for a prescribed time interval.

Pose estimation of non-cooperative satellites has been a hot topic in the study of astronautics as the visual feedback will highly enhance the safety of on-orbit services. A stereo vision system is proposed in this paper. It works as an eye-to-hand vision camera in the final approach phase Based on circular feature extraction, a closed-form solution is presented. The position and orientation of the adapter ring can be figured out in real-time as well as the unknown radius. Neither additional sensors nor prior knowledge is required, and the orientation-duality problem has been solved. It works well on the partial ellipses and is robust to outliers, noise and occlusions. Experimental results on both synthetic and real images have demonstrated the effectiveness and efficiency of the proposed method.

Learning from demonstration (LfD) is an appealing method of helping robots learn new skills. Numerous papers have presented methods of LfD with good performance in robotics. However, complicated robot tasks that need to carefully regulate path planning strategies remain unanswered. Contact or non-contact constraints in specific robot tasks make the path planning problem more difficult, as the interaction between the robot and the environment is time-varying. In this paper, we focus on the path planning of complex robot tasks in the domain of LfD and give a novel perspective for classifying imitation learning and inverse reinforcement learning. This classification is based on constraints and obstacle avoidance. Finally, we summarize these methods and present promising directions for robot application and LfD theory.

This paper presents a relative position and attitude estimation method using consecutive point clouds without feature extraction. Using this method, the inaccurate state estimation problems for non-cooperative targets caused by the mismatched point pairs or the low tracking accuracy of point cloud features can be resolved. First, point cloud registration is carried out by the transformation of the covariance matrices of the point cloud between two adjacent frames. Meanwhile, the random sample consensus algorithm is employed to reject the mismatched point pairs. Then, pose-graph optimization is adopted to eliminate the accumulated errors of consecutive point cloud registration. Finally, an Extended Kalman Filter is designed to estimate the position, velocity, and angular velocity of the target. The experimental results show that the covariance matrix transform algorithm can achieve the point cloud registration for close roto-translational motions, and the target motion state can be estimated effectively and continuously.

针对激光雷达与相机联合使用遇到的点云稀疏、相机受环境光照影响失真等问题,提出一种基于点云中心的激光雷达与相机自动配准方法,避免了传统联合标定需要手动选择特征点以及连续采集多帧等问题.该方法在对点云与图像预处理后,利用平面法向量的一致性实现多标定板点云自动分割,提取标定板在激光坐标系和相机坐标下的点云;然后通过点云聚集迭代求解中心点,实现两个传感器标定板对应点云中心的粗配准;最终利用迭代最近点算法进行精配准,获得标定参数,完成联合标定.实测表明,在激光雷达误差±3 cm范围内,点云正确投影比例达到97.93％,可以有效获取高精度联合标定参数,满足空间环境对激光雷达和相机数据融合的要求.

End-effector tools with low weight, multiple degrees of freedom have significant application significance. At present, motor-driven robot operating tools generally have more complicated transmission systems as well as actuators, and the weight is also heavy. In this paper, a bolt-screwing tool based on a pneumatic slip ring structure is designed, which can realize two-DOF motion of clamping-releasing and rotating. The tool consists of a pneumatic slip ring with sealed structure and a cylinder driven gripper. This article also introduces the control strategy required to screw the bolt, and describes the configuration of the experiment. Finally, the bolt is screwed into the thread hole in the experiment to verify the reliability of the screw assembly tool and the effectiveness of the corresponding control scheme.

Purpose The purpose of this paper is to develop an easily implemented and practical stabilizing strategy for the hardware-in-the-loop (HIL) system. As the status of HIL system in the ground verification experiment for space equipment keeps rising, the stability problems introduced by high stiffness of industrial robot and discretization of the system need to be solved ungently. Thus, the study of the system stability is essential and significant. Design/methodology/approach To study the system stability, a mathematical model is built on the basis of control circle. And root-locus and 3D root-locus method are applied to the model to figure out the relationship between system stability and system parameters. Findings The mathematical model works well in describing the HIL system in the process of capturing free-floating targets, and the stabilizing strategy can be adopted to improve the system dynamic characteristic which meets the needs of the practical application. Originality/value A method named 3D root-locus is extended from traditional root-locus method. And the improved method graphically displays the stability of the system under the influence of multivariable. And the strategy that stabilize the system with elastic component has a strong feasible and promotional value.

A dexterous, light weight arm with high precision, high speed manoeuvrability has been built for playing ping-pong. It is based on the modular design concept. The 6 DOFs robot arm consists of two shoulder joints, two elbow joints and one wrist joint. Brushless DC motors and harmonic reducer with a large central hole are applied on the shoulder joints and the elbow joints. The wrist joints are characterized by a bevel gear differential mechanism and driven by two flat DC motors, through a harmonic reducer and timing belt arrangement. Each modular joint contains a joint torque sensor, a joint position sensor, a motor position sensor, a current sensor and a temperature sensor. In each joint, there is a Field Programmable Gate Array(FPGA) for communication and control of BLDC motors. The core of the arm controller is a Intel Core II processor with 2.0 GHz. A PCI board based on FPGA is used for the communication between the arm controller and joint electrical units. Experiment results show that the dexterous arm can play table tennis successfully.