Robotics and Computer-integrated Manufacturing最新文献

Generating model-free grasps in complex scattered scenes remains a challenging task. Most current methods adopt PointNet++ as the backbone to extract structural features, while the relative associations of geometry are underexplored, leading to non-optimal grasp prediction results. In this work, a parallelized graph-based pipeline is developed to solve the 7-DoF grasp pose generation problem with point cloud as input. Using the non-textured information of the grasping scene, the proposed pipeline simultaneously performs feature embedding and grasping location focusing in two branches, avoiding the mutual influence of the two learning processes. In the feature learning branch, the geometric features of the whole scene will be fully learned. In the location focusing branch, the high-value grasping locations on the surface of objects will be strategically selected. Using the learned graph features at these locations, the pipeline will eventually output refined grasping directions and widths in conjunction with local spatial features. To strengthen the positional features in the grasping problem, a graph convolution operator based on the positional attention mechanism is designed, and a graph residual network based on this operator is applied in two branches. The above pipeline abstracts the grasping location selection task from the main process of grasp generation, which lowers the learning difficulty while avoiding the performance degradation problem of deep graph networks. The established pipeline is evaluated on the GraspNet-1Billion dataset, demonstrating much better performance and stronger generalization capabilities than the benchmark approach. In robotic bin-picking experiments, the proposed method can effectively understand scattered grasping scenarios and grasp multiple types of unknown objects with a high success rate.

Automating fabric manipulation in garment manufacturing remains a challenging task due to the characteristics of limp sheet materials and the diversity of fabrics used. This paper introduces an adaptive and optimized robotic fabric handling system, designed to address these challenges. The system comprises an industrial robot, four needle grippers, and a novel adaptive gripper jig system capable of adjusting the positions of the grippers adaptively to accommodate the shape and material properties of the garment fabric parts. To do this, an in-depth analysis of fabric gripping characteristics—accounting for material properties, gripping position, and fabric deformation—is conducted. A two-stage machine learning model predicting fabric deflection and folding is established from the analyzed data. This model is then incorporated into a vision-guided algorithm that determines the optimal gripping points on garment parts using corresponding CAD data. In addition, the exact position of the target fabric part is swiftly recognized via an algorithm that maps the real-time captured images to the CAD-based shape information. The decision-making information—namely optimal gripping points and garment part position—are subsequently transmitted to the robotic system for automated fabric handling process. The performance of the developed algorithms was quantitatively evaluated, and the integrated robotic system verified to be capable of completing garment manufacturing automation by connecting the processes of automatic fabric cutting and sewing.

Owing to the advantages of good flexibility and low cost, robots are gradually replacing manual labor as an effective carrier for the grinding and polishing of aeroengine blades. However, the geometric features of blades are complex and diverse, and the contour accuracy and surface quality requirements are high, making the robotic grinding and polishing of blades still a challenging task. For this reason, this article first designs a new device by integrating different tools, which can achieve full-feature grinding and polishing of blades. Then, in order to improve the accuracy and stability of force tracking during the robotic grinding and polishing processes, a variable impedance control approach with simultaneous changes in stiffness and damping and parameter boundaries is proposed. Finally, the superiority of the proposed variable impedance control method is verified by comparative experiments on surface tracking. In addition, by combining the device with the variable impedance control method in the robotic grinding and polishing experiments of an aeroengine blade, their effectiveness in practical situations is confirmed.

This article focuses on improving the compliance, efficiency, and robustness of batch precision assembly of small-scale pegs flexibly absorbed by a suction cup. The main contribution is that a force–vision fusion fuzzy control method (FVFFC) is proposed to achieve precision assembly with unknown clearance or interference fit. Both visual and force features are designed to describe the state of the peg and hole with the deformation of the suction cup. Then, a force–vision fusion control framework is proposed, where the visual features dynamically modify the reference position of admittance control and guide compliant adjustment of the peg angles. Furthermore, based on theoretical analysis, two fuzzy logic inference modules are developed to estimate the contact state as well as either the clearance or interference amount between the peg and hole in order to adaptively tune the control parameters. Finally, sufficient experiments are conducted to demonstrate the superiority and robustness of the FVFFC method.

This study presents the Reconfigurable and Responsive Robot Manufacturing (R3M) architecture, a novel framework engineered to autonomously adapt to fluctuating product variants and demands within manufacturing environments. At the heart of R3M lies an integrated architecture that ensures a seamless data flow between critical modules, facilitated by an advanced communication platform. These modules are central to delivering a range of services crucial for operational efficiency. Key to the architecture is the incorporation of Automated Risk Assessment aligned with ISO-12100 standards, utilizing ROS2 Gazebo for the dynamic modification of robot skills in a plug-and-produce manner. The architecture's unique approach to requirements definition employs AutomationML (AML), enabling effective system integration and the consolidation of varied information sources. This is achieved through the innovative use of skill-based concepts and AML Class Libraries, enhancing the system's adaptability and integration within manufacturing settings. The narrative delves into the intricate descriptions of products, equipment, and processes within the AML framework, highlighting the strategic consideration of profitability in the product domain and distinguishing between atomic and composite skills in equipment characterization. The process domain serves as an invaluable knowledge repository, bridging the gap between high-level product demands and specific equipment capabilities via process patterns. The culmination of these elements within the R3M framework provides a versatile and scalable solution poised to revolutionize manufacturing processes. Empirical results underscore the architecture's robust perception abilities, with a particular focus on a real-world application in robotic lamination stacking, elucidating both the inherent challenges and the tangible outcomes of the R3M deployment.

In recent years, with the deepening integration of digital twins (DT) and the Industrial Internet of Things (IIoT), solutions based on digital twins have been widely applied in IIoT scenarios. However, most existing solutions tend to overlook the latency issue during the interaction between mobile devices, such as industrial mobile robots (IMR), and their DTs while in motion. Excessive interaction latency can directly impair the real-time response capability and decision accuracy of industrial mobile robots, and in severe cases, it may lead to the failure of intricate industrial tasks. In order to solve the above problems, we propose a digital twin dynamic migration method for industrial mobile robots. Firstly, we design and implement a STGCN-Transformer-based movement trajectory prediction method for IMR to predict the future movement trajectory of IMR and pre-migrate the DT of IMR to all intelligent gateways (IG) within the prediction range. Then, we design and implement a Proximal Policy Optimization-based DT migration time determination method for IMR and obtain the migration timing of DT under the premise of balancing the DT migration overhead, the load of the IG where the DT is deployed, the load of the IG where the DT is connected, and the communication delay between the IMR and the IG where the DT is deployed. Next, the DT of the IMR is migrated based on the IMR’s anticipated trajectory and optimal times for migration, with the objective of minimizing the interaction latency between the IMR and its DT. Finally, we conduct simulation experiments on the proposed method. Through theoretical and simulation experiments, it has been proven that the proposed method can effectively ensure the dynamic interaction delay between the IMR and its DT during the moving process, thereby enhancing the real-time responsiveness and decision precision of the IMR.

Point clouds are widely used to construct models of workpieces using 3D scanners, especially where high-quality robotic and automatic 3D scanning is required in industries and manufacturing. In recent years, Point Cloud Quality Assessment (PCQA) has garnered increasing attention as it provides quality scores for entire point clouds, addressing issues such as downsampling and compression distortions. However, current PCQA methods cannot provide specific and local quality scores, which are necessary to facilitate rescanning and recompletion in 3D robotic scanning. Additionally, 3D data produced by robotic view-planning algorithms are usually considered the final result, where PCQA is typically not involved. In this paper, we bridge the gap between PCQA methods and practical robotic 3D scanning. We propose a no-reference PCQA method that recognizes sparse regions during 3D scanning, providing both local and overall quality scores. Unlike traditional methods that primarily consider density as a key metric, our method assumes that an expected 3D scan will have a uniformly distributed point cloud on surfaces. We analyze the quality of points by using geometric information from surfaces fitted to these points, which are mapped to a 2D distribution based on specified distances and angles. We conducted experiments on various datasets, including both synthetic and public datasets, to evaluate the accuracy and robustness of our method. The results show that our method can represent the quality on surfaces more accurately and robustly than density calculation methods. Additionally, it outperforms most existing PCQA methods in scenarios of downsampling, which is a common challenge in high-quality 3D scanning applications. The performance of our quality enhancement experiments on practical 3D scanning, conducted towards the end of our study, demonstrates significant potential for real-world applications. The related code is released at https://github.com/leihui6/PCQA.

Virtual Reality (VR) and Augmented Reality (AR) technologies have been separately applied to enhance a wide range of design and manufacturing operations across various industries. Most VR applications are primarily focused on product prototyping and personnel training, whereas AR is commonly used to facilitate manual operations in manufacturing such as assembly and maintenance tasks. This study proposes a novel concept of real-time integration between VR and AR scenes to leverage their individual advantages. A prototyping system is implemented based on this concept to enhance workplace ergonomics in the overhead assembly of car bodies. An operator undergoes the assembly training at a virtual workstation within VR, while an ergonomics expert evaluates the operator's actions from a third-person viewpoint using AR. The expert can iteratively adjust the workstation setup, supported by automatic human posture recognition and biomechanical analysis, to reduce the risk of musculoskeletal injuries during the process. An exemplary case demonstrates the practical value of real-time collaborative applications between VR and AR in the context of human-centric smart manufacturing. A usability study is conducted to verify the prototyping system using both subjective and objective measures. Integrating various reality technologies serves as an effective approach to improving human well-being in the manufacturing environment.

In automated fiber placement (AFP), addressing the continuous motion planning challenge of redundant layup manipulators in complex environments, this paper proposes an offline redundancy optimization algorithm based on improved RRT* (Rapidly-exploring Random Trees). This algorithm maximizes the utilization of kinematic redundancy to derive smooth joint trajectories devoid of collisions and singularities. Firstly, the algorithm entails constructing a search map by eliminating joint configurations that violate constraints, and subsequently planning and optimizing the joint path by minimizing a multi-objective cost under the map constraint. Furthermore, several strategies are introduced to enhance RRT* for redundancy optimization. These strategies include a piecewise Gaussian sampling strategy (PGSS) to guide efficient tree growth within complex channels and enable joint sampling constrained by task coordinates. Additionally, the improved Steering and Local Optimization method are proposed to plan joint motion while considering intermediate task sequences. The effectiveness of the proposed algorithm is demonstrated in handling complex motion planning scenarios, such as layup involving complex path curves or dense obstacles. Experimental results validate the algorithm's capability to find feasible collision-free and singularity-free paths in relevant scenarios, provided such paths exist. Moreover, trajectory smoothness is optimized with increasing iterations.

In emerging manufacturing facilities, robots must enhance their flexibility. They are expected to perform complex jobs, showing different behaviors on the need, all within unstructured environments, and without requiring reprogramming or setup adjustments. To address this challenge, we introduce the A3CQP, a non-strict hierarchical Quadratic Programming (QP) controller. It seamlessly combines both motion and interaction functionalities, with priorities dynamically and autonomously adapted through a Reinforcement Learning-based adaptation module. This module utilizes the Asynchronous Advantage Actor–Critic algorithm (A3C) to ensure rapid convergence and stable training within continuous action and observation spaces. The experimental validation, involving a collaborative peg-in-hole assembly and the polishing of a wooden plate, demonstrates the effectiveness of the proposed solution in terms of its automatic adaptability, responsiveness, flexibility, and safety.