Robotics and Computer-integrated Manufacturing最新文献

This paper introduces a methodology for information modelling and analysis of physical manufacturing processes for digital twins (DTs). It aims to establish a comprehensive and fundamental understanding of manufacturing processes regarding the specific purpose of the DT. Through this methodology, information entities within the manufacturing process that can be represented in DTs, along with their essential attributes, are systematically identified. To achieve this, an information model is firstly proposed to define such entities, termed as representative information. The attributes and hierarchy of entities are formulated based on a requirements analysis of the DT lifecycle. An Integration Definition for Process Modelling 0 (IDEF0) model, Petri nets, and a literature-based identification process are applied to represent the manufacturing process’s workflow and identify information entities. Moreover, the relative importance of representing each information entity in a DT is evaluated by integrating domain-specific knowledge with the specific purpose of the DT. Three types of information analysis are suggested, each with its corresponding methods: empirical analysis, theoretical analysis, and experimental analysis. Specifically, this study explores the material extrusion (MEX) process of the Prusa i3 MK3 printer, resulting in an information model consisting of 128 entities including 21 components, 25 activities and 82 properties. These information entities and associated attributes provide a reference for selecting and synchronizing specific physical information in a DT for estimating dimensional accuracy during the MEX process.

Machining quality prediction of multi-feature parts has been a challenging problem because of small dataset and inconsistent quality data distribution with respect to each machining feature. Transfer learning that leverages knowledge of one task and can be repurposed on another task seems a good solution for this purpose. However, traditional transfer learning typically has a single source domain and a target domain, which limits its applications in multi-source scenarios (e.g., multi-feature). To solve this issue, this paper proposes a novel integrated multi-source domain dynamic adaptive transfer learning (IMD-DATL) framework for machining quality prediction of multi-feature part machining systems. Specifically, a domain-sample similarity double matching multi-source domain integration method is designed to construct the integration knowledge transfer from multiple source domains to the target domain. A residual feature extraction network based on sample entropy-dynamic channel double-layer attention structure and a fine-grained transferable feature attention module are designed. These three attentions are used to improve the feature learning ability and the adaptation level to the predicted object in the three dimensions of sample, channel and data feature. Finally, multiple sets of comparative experiments in thin-walled part machining systems confirm the effectiveness and superiority of the proposed method for cross-domain quality prediction. Compared with other traditional transfer learning methods, the MAE, RMSE and Score on average of this method are increased by 5.47 %, 4.59 % and 4.84 %, respectively, compared with other multi-source domain adaptation methods, the MAE, RMSE and Score on average of this method are increased by 7.13 %, 7.37 % and 6.52 %, respectively.

The proliferation of technologies, such as extended reality (XR), has increased the demand for high-quality three-dimensional (3D) graphical representations. Industrial 3D applications encompass computer-aided design (CAD), finite element analysis (FEA), scanning, and robotics. However, current methods employed for industrial 3D representations suffer from high implementation costs and reliance on manual human input for accurate 3D modeling. To address these challenges, neural radiance fields (NeRFs) have emerged as a promising approach for learning 3D scene representations based on provided training 2D images. Despite a growing interest in NeRFs, their potential applications in various industrial subdomains are still unexplored. In this paper, we deliver a comprehensive examination of NeRF industrial applications while also providing direction for future research endeavors. We also present a series of proof-of-concept experiments that demonstrate the potential of NeRFs in the industrial domain. These experiments include NeRF-based video compression techniques and using NeRFs for 3D motion estimation in the context of collision avoidance. In the video compression experiment, our results show compression savings up to 48% and 74% for resolutions of 1920x1080 and 300x168, respectively. The motion estimation experiment used a 3D animation of a robotic arm to train Dynamic-NeRF (D-NeRF) and achieved an average peak signal-to-noise ratio (PSNR) of disparity map with the value of 23 dB and a structural similarity index measure (SSIM) 0.97.

Continuum robots have become more popular recently due to their scalable dexterity and mobility. However, they may suffer from issues like insufficient stiffness because they are designed to promote their flexibility. To address this issue and further improve their performance in all different application scenarios, stiffness flexibility is essential for this type of robot. Therefore, it is necessary to integrate stiffening techniques into both their mechanical structure and actuation approaches when developing continuum robots. To this end, it is crucial to explore how different stiffening approaches can be applied to various types of continuum robots across diverse applications. The primary goal of this survey paper is to provide a comprehensive review of the state-of-the-art research on stiffening techniques for continuum robots over the last two decades. We thoroughly analyse key techniques related to stiffness tunability mechanisms and stiffening methods. Additionally, we categorise these stiffening approaches on the basis of their properties and seek to understand the factors that limit their performance. This survey paper aims to assist robotic engineers in selecting appropriate stiffening techniques when designing continuum robots and serve as a basis for developing potential next-generation stiffening mechanisms.

There is a growing interest in implementing artificial intelligence for operations research in the industrial environment. While numerous classic operations research solvers ensure optimal solutions, they often struggle with real-time dynamic objectives and environments, such as dynamic routing problems, which require periodic algorithmic recalibration. To deal with dynamic environments, deep reinforcement learning has shown great potential with its capability as a self-learning and optimizing mechanism. However, the real-world applications of reinforcement learning are relatively limited due to lengthy training time and inefficiency in high-dimensional state spaces. In this study, we introduce two methods to enhance reinforcement learning for dynamic routing optimization. The first method involves transferring knowledge from classic operations research solvers to reinforcement learning during training, which accelerates exploration and reduces lengthy training time. The second method uses a state–space decomposer to transform the high-dimensional state space into a low-dimensional latent space, which allows the reinforcement learning agent to learn efficiently in the latent space. Lastly, we demonstrate the applicability of our approach in an industrial application of an automated welding process, where our approach identifies the shortest welding pathway of an industrial robotic arm to weld a set of dynamically changing target nodes, poses and sizes. The suggested method cuts computation time by 25% to 50% compared to classic routing algorithms.

Task planning and action planning for workshop machines are essential for modern manufacturing. Traditionally, these two problems are solved independently with elaborate manual methods. However, personalized customization introduces more dynamic exogenous events into the manufacturing system, and it is then impossible to consider all possible dynamic scenarios manually. This paper focuses on online automated planning, generating new plans automatically in response to new dynamic situations. We first formulate the planning problem for a flexible manufacturing system as a fully observable nondeterministic planning problem. Second, a hierarchical automated online planning approach is presented. Finally, the effectiveness of the proposed approach is verified by an ARIAC 2022 competition environment.

The Cloud Manufacturing Service Composition and Scheduling (CMfg-SCS) are essential processes in cloud manufacturing. Flexible Manufacturing Services (FMS), such as those provided by industrial robots, are widely used in cloud manufacturing to improve service quality and efficiency. Traditional CMfg-SCS methodologies, however, fall short in effectively managing the inherent temporal-dynamic QoS and flexible capability of FMS. To overcome these challenges, we propose a novel Cloud Manufacturing Service Cloud-edge Collaboration Composition and Scheduling (CMfg-SCCCS) method for FMS. Firstly, the service-task matching hypernetwork is constructed, and the temporal-dynamic QoS and flexible capacity of FMS are modeled. Subsequently, we develop a CMfg-SCCCS optimization model aimed at three objectives, along with a cloud-edge collaboration scheduling mechanism to harmonize cloud and edge-local tasks. Finally, a multi-population co-evolution algorithm with adaptive meta-knowledge transfer mechanism is proposed to solve the complex optimization model. The computational experiments serve to validate the effectiveness of the CMfg-SCCCS method and further reveal the superiority of the co-evolution algorithm in enhancing both the convergence and diversity of the population.

Elasto-geometrical calibration is crucial for enhancing the absolute accuracy of robots in machining operations through the identification and compensation of parameter errors. However, the presence of inconsistent measurement units and improper selection of measuring poses can result in the ill-conditioned identification matrix (ICIM) issue, consequently impacting the accuracy of parameter identification. This paper introduces a novel method to tackle this challenge. Initially, an elasto-geometrical error model is developed based on the orientation-independent measurements (OIM), efficiently reducing the impact of mismatched positions and orientations on the ICIM problem. Subsequently, a PSO-SFFS algorithm is proposed to optimize the measurement configurations and minimize the influence of measurement noise. Furthermore, the incorporation of screw theory and the consideration of parallelogram mechanisms enhance the precision and comprehensiveness of the error model. Subsequent to the development of the error model, calibration comparison experiments are conducted on an industrial robot. Both simulation and experimental results validate the effectiveness of the proposed method in improving parameter identification accuracy.

Compared to a single robot, multi-robot systems (MRS) offer several advantages in complex multi-task scenarios. The overall efficiency of MRS relies heavily on an efficient task allocation and scheduling process. Multi-robot task allocation (MRTA) is often formulated as a multiple traveling salesman problem, which is NP-hard and typically addressed offline. This paper specifically addresses the online allocation problem in multi-manipulator systems within multi-task scenarios. The tasks are initially pre-allocated to alleviate the computational burden of online allocation. Subsequently, considering collision constraints, we search for the current feasible set of manipulators and employ greedy algorithms to achieve local optima as the online allocation result within this set. Our method can handle the online addition of new, unknown tasks to the task list. Moreover, we demonstrate the feasibility of our approach through simulations and on a realistic platform, where multiple manipulators are tasked with polishing the white body of automobile parts. The results demonstrate that our method is effective and efficient for online allocation and scheduling scenarios.

The complex background on the car body surface, such as the orange peel-like texture and shiny metallic powder, poses a considerable challenge to automated defect detection. Two mainstream methods are currently used to tackle this challenge: global information-based and attention mechanism-based methods. However, these methods lack the capability to integrate valuable global-to-local information and explore deeper distinguishable features, thereby affecting the overall detection performance. To address this issue, we propose a novel attention enhanced global–local refined detection network (AEGLR-Net), which can perform effective global-to-local refined feature extraction and fusion. First, we design an adaptive Transformer–CNN tandem backbone (ATCT-backbone) to dynamically aware valuable global information and integrate local details to comprehensively extract specific features between defects and complex backgrounds. Then, we propose a novel refined cross-dimensional aggregation (RCDA) attention to facilitate the point-to-point interaction of multidimensional information, effectively emphasizing the representation of deeper discriminative defect features. Finally, we construct an attention-embedded flexible feature pyramid network (AE-FFPN), which incorporates the RCDA attention to guide the feature pyramid network in targeted feature fusion, thereby enhancing the efficiency of feature fusion in the detection model. Extensive comparative experiments demonstrate that the AEGLR-Net outperforms state-of-the-art approaches, attaining exceptional performance with 89.2 % mAP (mean average precision) and 85.5 FPS (frames per second).