Complete, physically consistent dynamics parameters are required for forward dynamics simulation and model-based control methods that explicitly use a separated generalized mass matrix. Complete dynamics parameters are not provided by the manufacturer. The available CAD models contain neither closed surfaces (and thus no volumetric information) nor density information from which the required mass properties could be determined reliably. This paper reports complete reference parameter sets for the Universal Robots UR5e and UR10e using an established physically consistent identification approach. To obtain representative results, two UR5e and four UR10e robots are identified, and both robot-specific and averaged parameter sets are reported. The identified inertia parameters agree well among the robots, whereas the friction parameters vary. For the evaluated UR10e trajectory, the torque differences between the robot-specific and averaged models remain within approximately 5% of the maximum joint torques, supporting the use of the averaged sets as nominal reference parameters. A forward dynamics simulation further demonstrates the use of the complete parameter sets in model-based control. For the evaluated slow pick-and-place motion, the Coriolis and centrifugal torque contributions are sufficiently small to be neglected. The computational benefit is assessed separately: omitting these terms reduces the mean model evaluation time from 8.9 μs to 3.3 μs per sample. Although the averaged parameter sets provide useful nominal models, robot-specific friction identification may still be required depending on the specific task. Practical details relevant to reproducing the identification are also discussed. This paper should serve the robotics community as a reliable and representative reference for physically consistent dynamics parameters of the UR5e and UR10e robots.

Abstract This article proposes a design optimization strategy aimed at improving the performance of a linear belt drive system. The elasticity of the belt and nonlinear disturbances causes undesirable vibrations and reduces the trajectory tracking accuracy, affecting the precise positioning of the drive and reducing its efficiency. These issues are addressed by developing kinematics and dynamics models of the drive, which include position-dependent stiffness, damping, and nonlinear frictions. Based on these models, design optimization and control strategies are devised to enhance the efficiency of the system. To analyze and further enhance the performance of the system under actual operating conditions, a virtual prototype (VP) model of this drive is developed using the multi-body dynamics simulations tool. This technique reduces the time and cost of prototyping by enabling design iterations without the need for physical prototyping. The results obtained from the VP and analytical modeling are validated experimentally to ensure their accuracy and effectiveness in capturing the behavior of the drive. In the end, a reliable and cost-effective solution is provided for designing a high performance linear belt drive system.

Convolutional Kolmogorov–Arnold Networks (KANs) replace the fixed weights of a convolutional kernel with learnable univariate functions. The dominant formulation attaches one such function to every kernel entry and lets it act on pixel values, expressive but parameter-heavy and prone to overfitting. We argue that the learnable functions are better placed in the structure of the convolution than on each edge, and we organise the design space along a single axis: whether the function acts on the pixel values or on the filter shape. We study three realisations. SV-KAN applies one shared univariate function to the values and leaves the spatial filter free and static, aa classical convolution with a single learnable shared activation. AG-KAN keeps the shared value function but supplies the spatial structure through a content-adaptive Gaussian gate. RF-KAN instead moves the learnable functions onto the filter shape, building each filter from oriented ridge profiles expanded in a localised oscillatory (Morlet) wavelet basis with content-adaptive amplitudes. Under a matched four-layer protocol with in-run references and three seeds, RF-KAN and SV-KAN reach 88.47±0.10% and 88.20±0.31% on CIFAR-10 and 64.40±0.19% and 64.57±0.30% on CIFAR-100, at about 0.4M parameters. At this matched scale the shape model and the simplest value model meet at the top, both above a plain convolution and every per-edge KAN we tested, including the official Gram variant, at roughly a fifth of the parameters. A controlled study attributes the RF-KAN gain to an intrinsically localised oscillatory basis and to content adaptivity, and an ablation that removes the learned shape entirely, leaving only the shared value function, collapses accuracy by over forty points, identifying the learned shape as the load-bearing ingredient at this scale.

This paper pioneers the integration of hyperspectral imaging and robotics for the automated analysis of cultural heritage, representing a measurable advancement over existing manually operated systems. For the first time in the cultural heritage domain, a compact push-broom hyperspectral camera working in the VNIR range has been successfully mounted on a robotic arm, enabling precise and repeatable acquisition trajectories without the need for manual intervention. Unlike traditional approaches that rely on fixed paths or manual repositioning, the proposed approach allows dynamic and programmable imaging of both planar and volumetric objects, greatly improving adaptability to complex geometries. The integrated system achieves spectral reliability comparable to established manual methods, while offering superior flexibility and scalability. Current limitations, particularly regarding the illumination setup, are discussed alongside planned optimisation strategies.

The inspection and maintenance of pressure equipment present significant technical and safety challenges yet is a critical stage in its preventive maintenance. Traditional visual inspection methods require skilled operators and involve prolonged downtime, increasing operational costs and risk exposure. This paper reviews 90 key studies on inspection technologies for pressure equipment from 1989 to 2024, evaluating their advantages over traditional inspection, such as reliance on highly skilled operators, prolonged inspection durations, and significant safety risks. This study highlights current technological advancements—such as the use of several non-destructive testing techniques, robotics, and artificial intelligence—identifies barriers to widespread implementation, and discusses future research directions to improve robotic solutions for pressure equipment inspection. Automated robotic systems offer a promising alternative to enhance safety. However, the adoption of these technologies remains limited due to regulatory challenges, accessibility constraints, and the complexity of internal structures. The paper underscores the critical role of continued research and innovation in robotic technologies to bridge these gaps and accelerate adoption. Finally, an automated robotic solution is presented, integrating cognitive mechatronics and artificial intelligence with advanced sensors for defect detection and inspection inside pressure equipment. Advancing automation in this field is crucial for enhancing efficiency, reducing economic impact, and ensuring safer maintenance operations in dangerous environments.

High-fidelity 3D scanning is essential for preserving cultural heritage artefacts, supporting documentation, analysis, and long-term conservation. However, conventional methods typically require specialized expertise and manual intervention to maintain optimal scanning conditions and coverage. We present an automated two-robot scanning system that eliminates the need for handheld or semi-automatic workflows by combining coordinated robotic manipulation with high-resolution 3D scanning. Our system parameterizes the scanning space into distinct regions, enabling coordinated motion planning between a scanner-equipped robot and a tray-handling robot. Optimized trajectory planning and waypoint distribution ensure comprehensive surface coverage, minimize occlusions, and balance reconstruction accuracy with system efficiency. Experimental results show that our approach achieves significantly lower Chamfer Distance and higher F-score compared to baseline methods, offering superior geometric accuracy, improved digitization efficiency, and reduced reliance on expert operators.

This paper introduces an innovative robotic solution to autonomously plan stitching paths for two fabric layers. The developed system is inspired by the conventional manufacturing process and aims at assisting the human effort while ensuring high-quality results. Extensive experiments reveal comparable performance to skilled human operators in terms of path accuracy and stitching quality, accommodating various fabric types and patterns. This research advances robotic fabric processing offering a modular and flexible solution. It holds promise for the textile industry and precision-dependent applications, simplifying automation and enhancing efficiency.

Among the many challenges of parallel kinematic manipulators, achieving high-speed and accurate control remains crucial. Estimating their dynamic properties is essential for designing precise and efficient control schemes. Conventional methods for dynamic model identification have been effective, though deep learning approaches have historically faced limitations due to data inefficiencies. However, recent advancements in physics-informed neural networks (PINNs) offer a way to improve both control and the extraction of interpretable physical properties from these robots. In this work, we propose and validate a PINN-based dynamic model for a Delta parallel robot, specifically the ABB IRB 360-6/1600. Our approach incorporates known physical properties, such as mass matrix sparsity, to improve accuracy and computational efficiency in dynamic model identification. To the best of our knowledge, this is the first study applying PINNs to model parallel robots. The method is validated experimentally, and its performance is compared to a validated identification technique for physically consistent identification, demonstrating the effectiveness of this approach for real-world applications in parallel robots.

The use of two-dimensional (2D) images to extract information about the three-dimensional (3D) structure of complex objects is a widely adopted approach. However, it often requires capturing a large number of images and entails significant computational costs. In order to address these challenges, this paper presents an optimization method for determining the optimal placement of viewpoints in 3D reconstruction, improving efficiency while maintaining high accuracy. The algorithm is designed to facilitate object manipulation by robotic systems, particularly for identifying optimal grasping points. The experimental setup includes a transparent surface where a set of 5 to 9 objects is arranged for scanning. To simulate the workspace of a robotic arm, the algorithm selects the best viewpoints from a discrete set of possible camera placements, positioned at predefined distances along the parallels of a sphere enclosing the tray that holds the objects. To ensure computational efficiency, the algorithm employs a greedy approach, enabling fast calculations while maintaining high precision in viewpoint selection. Its effectiveness is evaluated through several numerical tests, assessing performance under different conditions by varying the number of objects and comparing visibility results when considering either the entire object surface or only the external surface. Furthermore, this study introduces an innovative feature that excludes internal object regions from the visibility computation, enhancing performance in complex environments. Experimental results demonstrate that the method achieves high surface visibility even with a limited number of viewpoints, particularly when the object density within the tray is not excessively high.

A comprehensive literature review on the kinematics and dynamics modeling and virtual prototyping (V.P) of the Cartesian robots with a flexible configuration is presented in this paper. Different modeling approaches of the main components of the Cartesian robot, which includes linear belt drives and structural components, are presented and discussed in this paper. Furthermore, the vibrations modeling, trajectory planning, and control strategies of the Cartesian robot are also presented. The performance optimization of the Cartesian robot is discussed here, which is affected by the highly flexible configuration of the robot incurred due to high-mix, low-volume production. The importance of virtual prototyping techniques, like finite element analysis and multi-body dynamics, for modeling Cartesian robots or its components is presented. Design and performance optimization methods for robots with a flexible configuration are discussed, although their application to Cartesian robots is rare in the literature and it presents an exciting opportunity for future research in this area. This review paper focuses on the importance of further research on the virtual prototyping tools for flexibly configured robots and their integration with experimental validation. The findings offer useful insights to industries looking to maximize their production processes while keeping the customization, reliability, and efficiency.

Model-based control is crucial for efficient robotic operations, yet accurately identifying robot dynamics remains challenging, particularly for parallel kinematic manipulators (PKMs). This work leverages physics-informed neural networks (PINNs), specifically the Deep Lagrangian Network in combination with non-symmetric Coulomb friction, to achieve physically consistent dynamics models by incorporating principles such as energy conservation and friction modeling. Validated on the ABB IRB 360-6/1600 Delta robot, the approach demonstrates high fidelity in torque prediction and effective real-time control implementation on industrial hardware under stringent computational constraints. Experimental results highlight improved torque prediction accuracy, reduced trajectory tracking lag, and robust handling of complex dynamic interactions, paving the way for adaptive and efficient industrial automation.

In the world, there are hundreds of automation devices and equipment on the bridges for automatic inspection and/or monitoring. These range from simple travellers to highly complex robots, all of which require infrastructure such as rails, cables, beams, and tracks to operate safely and autonomously. Moreover, if the design of inspection robots is a case-specific compromise between competing needs for sophisticated inspection sensing and for flexible locomotion in challenging field environments, that infrastructure could be even more complex. This complexity necessitates creating a contact point between the civil structure and the advanced device, bridging two domains with significantly different measurement tolerances. This detail can impact the movement accuracy of mobile systems, transforming a structured environment into an unstructured one. Consequently, this can lead to overly complicated navigation controls or incorrect positioning of the inspection probes. To address these limitations, either a human operator must remain involved, which reduces the benefits of automation, or innovative solutions must be introduced in autonomous systems. Therefore, this work aims to define the most influential parameters and provide potential solutions to improve the efficiency and accuracy of infrastructure inspection and maintenance.

Manipulating soft materials has always been one of the most difficult problems in robotics, due to the non-linear mechanical behaviour of fabrics. Therefore, the automation of systems based on the manipulation of soft materials (such as the clothing industry) has been very limited. We present a robotic cell that supports workers by automating the production of cyclist garments, composed of an elastic cloth and a foam pad to sew together. The robotic cell is comprised of two robotic arms equipped with a two-finger parallel gripper and a pneumatic needle gripper to flatten the cloth and pick the foam pad, respectively. Moreover, a Cartesian robot is employed to drive the two fabrics under the needle of a sewing machine. This project aims to improve the productivity of garments and the working conditions of the operators. The results obtained by the robotic cell are comparable with conventional ones both in quality and production time. In addition, the modularity underlying the design of this structure ensures a high degree of flexibility. Therefore, the system can be used to make all types of garments.

Pose estimation of cultural heritage artifacts is essential for the accurate alignment of 3D scans. This paper proposes a novel robotic arm-based strategy named Automated Artifacts Position and Orientation Estimation (AAPOE), which detects artifacts, re-identifies them if moved, and estimates and tracks their poses within the workspace. The robotic arm is equipped with a 3D scanner that captures depth maps from different positions, converting them into point clouds to form a complete scene. AAPOE filters out the ground plane, clusters the point cloud to detect individual artifacts, and re-identifies them if moved. The artifacts detected for the first time are processed to assign unique poses to each of them based on principal component analysis of their point clouds. When the artifacts are re-identified, AAPOE applies a rotate-until-converge approach for alignment, ultimately estimating their poses. These steps ensure accurate pose estimation and tracking, which are critical for fine-grained scanning and robot manipulation in cultural heritage applications. Experiments demonstrate that the estimated poses accurately align the artifacts placed and scanned in any random position within the workspace.

The robotics for Inspection and Maintenance (I&M) was introduced decades ago. Increasing their technologies, they were more and more used in more complex scenarios, thanks also to the legged robots able to move on uneven terrain as well as drones always more autonomous. Energy production plants were investigated in this paper, to evaluate how is the suitability for autonomous I & M; that's matter of facts, that, until now, robotic based inspection of photovoltaic systems was considered only as one of the possible I&M strategies feasible for this category of energy systems. In this work, the advantages of autonomous inspection systems for large photovoltaic systems are discussed by connecting them to the specific characteristics of a large photovoltaic utility plant. The objective is to highlight that, based on a large investigation, in the case of photovoltaic energy, robotic inspection is a necessity rather than an option. [GRAPHICS] .

This paper investigates the application of mobile robotic platforms for visual data capture in infrastructure inspection tasks. The captured data offer significant value for both manual and automated inspection processes. It can produce detailed visual information for human inspectors and serve as input for automated systems to detect anomalies or assist inspectors through computer-aided analysis. Additionally, these data can be integrated into the robot navigation system for real-time path optimisation. A critical challenge in optimising data capture is highlighted: balancing the desired precision with the time invested in inspections. The study explores this trade-off by analysing the impact of motion blur on measurement errors. Capturing high-quality images with minimal motion blur necessitates slower inspection speeds. The findings suggest that for extensive inspection areas, prioritising mid-range object distances can optimise data capture, as errors increase at a slower pace at these distances compared to closer or farther ranges. This research paves the way for further advancements. Future areas of exploration include evaluating noise reduction techniques, incorporating real-world complexities into testing environments, and investigating the impact of capture speed on machine learning algorithms.

The determination of the dynamic properties of a robot is especially important for designing highly accurate and efficient control systems. Conventional methods for dynamic model identification have proven to be effective, where deep learning (DL) approaches have shown limits due to data inefficiencies. However, thanks to novel physics-informed DL architectures, such as Deep Lagrangian Networks (DeLaN) [1], it is possible to control and extract interpretable physical information of a robot. This paper introduces an augmented DeLaN architecture for linear viscous and non-symmetrical Coulomb friction identification, which also learns motor parameters such as rotor inertia. An approach is proposed for comparing this method with the conventional dynamic identification and previous DeLaN implementations. Moreover, our friction and rotor inertia identification is validated, and the performance of our model is analyzed with a real robot (UR5e).

The inspection of archaeological sites is a multifaceted work that combines technology, tradition and interdisciplinary knowledge. In this context, we present a robotic system designed to autonomously detect defects and damages on the walls of ancient buildings that make up the Pompeii Archaeological Park. Our system consists of an autonomous rover (RINGHIO: Robot for Inspection and Navigation to Generate Heritage and Infrastructures Observation) equipped with a vision system mounted on a vibration compensation device (DDB: Defect Detection Box). We have conducted rigorous tests on a carefully selected section of the ancient road leading to Porta Stabia and an insula strategically chosen due to various building wall defects and complex topography. This study highlights the fundamental role of automated survey techniques in the preservation of our shared cultural heritage. The preservation of archaeological sites is of great importance and this work demonstrates the importance of collaborative efforts to ensure the longevity of these priceless monuments for future generations.

This study introduces a method based on the Asynchronous Kalman Filter (AKF) for accurately estimating the state of rigid-flexible robotic manipulators, tackling the integration of a flexible link with a rigid manipulator. Conventional state estimation techniques struggle with the dynamic complexities of such systems and the challenge of multi-rate sensor measurements. The proposed AKF-based approach significantly improves estimation accuracy and system performance by asynchronously fusing measurements from an IMU sensor and UR5e arm and correcting them using visual feedback. The research evaluated four distinct methodologies within the experimental framework: (i) reliance solely on IMU acceleration data, (ii) an AKF excluding camera data, (iii) an AKF incorporating camera measurements, and (iv) the proposed AKF-based approach being corrected by the camera. The experimental results have highlighted the enhanced effectiveness of the proposed method in estimating the state of rigid-flexible link manipulators compared to the alternatives.