Mechatronics最新文献

{"title":"Testing human-hand segmentation on in-distribution and out-of-distribution data in human–robot interactions using a deep ensemble model","authors":"Reza Jalayer ,&nbsp;Yuxin Chen ,&nbsp;Masoud Jalayer ,&nbsp;Carlotta Orsenigo ,&nbsp;Masayoshi Tomizuka","doi":"10.1016/j.mechatronics.2025.103365","DOIUrl":"10.1016/j.mechatronics.2025.103365","url":null,"abstract":"<div><div>Reliable detection and segmentation of human hands are critical for enhancing safety and facilitating advanced interactions in human–robot collaboration. Current research predominantly evaluates hand segmentation under in-distribution (ID) data, which reflects the training data of deep learning (DL) models. However, this approach fails to address out-of-distribution (OOD) scenarios that often arise in real-world human–robot interactions. In this work, we make three key contributions: first we assess the generalization of deep learning (DL) models for hand segmentation under both ID and OOD scenarios, utilizing a newly collected industrial dataset that captures a wide range of real-world conditions including simple and cluttered backgrounds with industrial tools, varying numbers of hands (0 to 4), gloves, rare gestures, and motion blur. Our second contribution is considering both egocentric and static viewpoints. We evaluated the models trained on four datasets, i.e. EgoHands, Ego2Hands (egocentric mobile camera), HADR, and HAGS (static fixed viewpoint) by testing them with both egocentric (head-mounted) and static cameras, enabling robustness evaluation from multiple points of view. Our third contribution is introducing an uncertainty analysis pipeline based on the predictive entropy of predicted hand pixels. This procedure enables flagging unreliable segmentation outputs by applying thresholds established in the validation phase. This enables automatic identification and filtering of untrustworthy predictions, significantly improving segmentation reliability in OOD scenarios. For segmentation, we used a deep ensemble model composed of UNet and RefineNet as base learners. Our experiments demonstrate that models trained on industrial datasets (HADR, HAGS) outperform those trained on non-industrial datasets, both in segmentation accuracy and in their ability to flag unreliable outputs via uncertainty estimation. These findings underscore the necessity of domain-specific training data and show that our uncertainty analysis pipeline can provide a practical safety layer for real-world deployment.</div></div>","PeriodicalId":49842,"journal":{"name":"Mechatronics","volume":"110 ","pages":"Article 103365"},"PeriodicalIF":3.1,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144329530","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Matrix-sensitivity-based active disturbance rejection control for hydraulic servo positioning systems with friction compensation","authors":"Kaixian Ba ,&nbsp;Ning Liu ,&nbsp;Jinbo She ,&nbsp;Yuan Wang ,&nbsp;Guoliang Ma ,&nbsp;Bin Yu ,&nbsp;Xiangdong Kong","doi":"10.1016/j.mechatronics.2025.103378","DOIUrl":"10.1016/j.mechatronics.2025.103378","url":null,"abstract":"<div><div>Accurate position regulation in hydraulic servo systems (HDU) plays a critical role in ensuring system stability, operational efficiency, and achieving high-accuracy performance. However, friction-induced nonlinearities, including Stribeck effects and internal friction dynamics, significantly impact tracking accuracy. This paper introduces a matrix-sensitivity-based active disturbance rejection control (MSADRC) method that compensates for friction without requiring an explicit friction model. By leveraging matrix sensitivity, MSADRC effectively decouples system dynamics and enhances control accuracy, particularly in suppressing frictional effects. A third-order extended state observer (ESO) first estimates total system disturbances, while a model predictive mechanism converts nonlinear time-varying disturbances into a feedforward compensation term. The resulting matrix sensitivity-based compensation optimally adjusts system response, ensuring improved performance. Experimental results show that MSADRC effectively mitigates nonlinear disturbances, reducing peak error by up to 55 % compared to conventional ADRC methods. This approach provides a reliable and efficient strategy to address adaptive friction compensation issues in hydraulic control systems.</div></div>","PeriodicalId":49842,"journal":{"name":"Mechatronics","volume":"110 ","pages":"Article 103378"},"PeriodicalIF":3.1,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144470701","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"3D coordinate sensing with nonsmooth friction dynamical discontinuities compensation in laser scanning system","authors":"Oleg Sergiyenko ,&nbsp;José A. Núñez-López ,&nbsp;Vera Tyrsa ,&nbsp;Rubén Alaniz-Plata ,&nbsp;Oscar M. Pérez-Landeros ,&nbsp;César Selpúlveda-Valdez ,&nbsp;Wendy Flores-Fuentes ,&nbsp;Julio C. Rodríguez-Quiñonez ,&nbsp;Fabian N. Murrieta-Rico ,&nbsp;Vladimir Kartashov ,&nbsp;Marina Kolendovska","doi":"10.1016/j.mechatronics.2025.103382","DOIUrl":"10.1016/j.mechatronics.2025.103382","url":null,"abstract":"<div><div>This study aimed to address the issue of laser ray spatial positioning to mitigate discontinuities in the dynamics caused by nonsmooth friction effects by the direct application of control theory with improved friction compensation. Analyzing physical phenomena on micro-relieved surfaces through SEM methods, the obtained data about surface characteristics helps synthesize a corresponding control law for the laser positioner and conduct its stability analysis. This work considers a patented laser scanning system incorporating a laser positioning mechanism with inherent friction. SEM micrograph analysis of the friction zone was conducted to compare microscopic imperfections of steel surfaces, which helped infer the dynamics of an internal variable ‘z’ in the friction model and determine a reference value for control synthesis. A nonlinear control algorithm was proposed to compensate for friction to enhance positioning accuracy. The global asymptotic stability of the system was proven using Lyapunov’s direct method and Barbalat’s lemma. Experimental implementation on an STM32 board demonstrated a significant reduction in the uncertainty associated with sensing 3D coordinates using the friction-compensated laser scanning system.</div></div>","PeriodicalId":49842,"journal":{"name":"Mechatronics","volume":"110 ","pages":"Article 103382"},"PeriodicalIF":3.1,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144588654","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Proving the stability of cycle navigation using capture sets","authors":"Quentin Brateau,&nbsp;Loïck Degorre,&nbsp;Fabrice Le Bars,&nbsp;Luc Jaulin","doi":"10.1016/j.mechatronics.2025.103385","DOIUrl":"10.1016/j.mechatronics.2025.103385","url":null,"abstract":"<div><div>Navigating Autonomous Underwater Vehicles (AUVs) presents significant challenges due to the absence of traditional localization systems. Cycle navigation emerges as a promising paradigm, enabling reliable navigation using minimal exteroceptive measurements. This approach leverages predefined cyclic trajectories, which are stabilized based on environmental feedback, ensuring frugal and discreet operations without reliance on high computational power or extensive sensor systems. This work aims to prove the stability of the cycle navigation. As cycle navigation is a non-linear system governed by a discrete inclusion condition, conventional methods have trouble to prove its stability. For this reason, this paper focuses on set methods to prove the stability of cycle navigation. The stability is proven by exhibiting a positive invariant set, which is a set stable by application of the evolution function of the system. This ensures that the evolution function will not remove states from the positively invariant set. Then, the characterization of the capture basin is an asset when performing cycle navigation, as it represents the set of initial states for the system which leads to the positive invariant set. Once the system reaches either the capture basin or the positive invariant set, which are generalized as a capture set, it remains captured forever. This approach not only guarantees the stability of the system in the neighborhood of the equilibrium point, but also establishes that it exists an area in which the stability of the cycle navigation will lead to a stable behavior. This work offers a robust, computationally efficient alternative to traditional stability methods, particularly suited for resource-constrained AUVs, because the underwater environment lacks suitable, cheap and easy-to-use localization methods, which forces us finding alternative ways to navigate and explore this particular environment.</div></div>","PeriodicalId":49842,"journal":{"name":"Mechatronics","volume":"110 ","pages":"Article 103385"},"PeriodicalIF":3.1,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144739056","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Comfort-enhanced longitudinal control for DDEVs: A robust brake coordination approach leveraging reactive anti-dive forces","authors":"Yanjun Ren ,&nbsp;Tong Shen ,&nbsp;Mingzhuo Zhao ,&nbsp;Fanxun Wang ,&nbsp;Liwei Xu ,&nbsp;Guodong Yin","doi":"10.1016/j.mechatronics.2025.103357","DOIUrl":"10.1016/j.mechatronics.2025.103357","url":null,"abstract":"<div><div>Distributed drive electric vehicles actuated by in-wheel motors and brake-by-wire systems enable tracking target motion while improving extra vehicle performance. Outboard brake torque allocated on front and rear wheels generates diverse vertically reactive anti-dive forces, providing an innovative approach to mitigate brake dive without requiring active suspensions. However, the differing dynamics of regenerative and hydraulic braking, along with multiple uncertain vehicle parameters, pose significant challenges to achieving robustness under mixed uncertainties. Moreover, pitch-induced bias in onboard acceleration measurements further degrades control accuracy. To address above problems, this paper proposes a robust, comfort-enhanced longitudinal control system with coordinated braking. A three-degree-of-freedom vehicle dynamics model is developed to incorporate the effect of anti-dive forces. For accurate feedback, a robust <span><math><mrow><msub><mrow><mi>H</mi></mrow><mrow><mn>2</mn></mrow></msub><mo>/</mo><msub><mrow><mi>H</mi></mrow><mrow><mi>∞</mi></mrow></msub></mrow></math></span> observer is designed to compensate pitch-variation-related acceleration measurement biases. By integrating dynamic and parametric uncertainties into the control-oriented model, the mixed <span><math><mi>μ</mi></math></span>-synthesis is employed to design a two-degree-of-freedom controller to robustly optimize the acceleration tracking and anti-dive performance. Compared to the controller designed by standard <span><math><mi>μ</mi></math></span>-synthesis, the proposed approach achieves a 10% improvement in robust performance. Real-vehicle experiments validate the system’s effectiveness, demonstrating over a 27% reduction in pitch angle while maintaining satisfactory acceleration responses under blended braking conditions.</div></div>","PeriodicalId":49842,"journal":{"name":"Mechatronics","volume":"110 ","pages":"Article 103357"},"PeriodicalIF":3.1,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144281075","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Active compensation of position dependent flexible dynamics in high-precision mechatronics","authors":"Yorick Broens ,&nbsp;Hans Butler ,&nbsp;Ramidin Kamidi ,&nbsp;Koen Verkerk ,&nbsp;Siep Weiland","doi":"10.1016/j.mechatronics.2025.103377","DOIUrl":"10.1016/j.mechatronics.2025.103377","url":null,"abstract":"<div><div>Growing demands in the semiconductor industry necessitate increasingly stringent requirements on throughput and positioning accuracy of lithographic equipment. Meeting these demands involves employing highly aggressive motion profiles, which introduce position-dependent flexible dynamics, thus compromising achievable position tracking performance. This paper introduces a control approach enabling active compensation of position-dependent flexible dynamics by extending the conventional rigid-body control structure to include active control of flexible dynamics. To facilitate real-time implementation of the control algorithm, appropriate position-dependent weighting functions are introduced, ensuring computationally efficient execution of the proposed approach. The efficacy of the proposed control design approach is demonstrated through experiments conducted on a state-of-the-art extreme ultraviolet (EUV) wafer stage.</div></div>","PeriodicalId":49842,"journal":{"name":"Mechatronics","volume":"110 ","pages":"Article 103377"},"PeriodicalIF":3.1,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144470703","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Natural efficient gaits from Nonholonomic Locomotion Nonlinear Normal Mode (NL-NNM): The Pendrivencar case","authors":"Mirado Rajaomarosata,&nbsp;Luc Jaulin,&nbsp;Lionel Lapierre,&nbsp;Simon Rohou","doi":"10.1016/j.mechatronics.2025.103366","DOIUrl":"10.1016/j.mechatronics.2025.103366","url":null,"abstract":"<div><div>Bio-inspired robots remain far less energy-efficient than animals because conventional controllers impose trajectories that fight passive dynamics, whereas animals exploit resonance through <em>natural nonlinear normal modes (NNM)</em>, whose periodic internal motions form a smooth 2D invariant surface; We ask how to define and compute the <em>natural motions of a conservative locomotion system</em>: propulsion arises only from <em>no-slip constraints</em>, and once initiated, a gait persists without actuation—like a frictionless pendulum. We tackle non-holonomic constraints on the <em>Pendrivencar</em>, a vehicle driven by a <em>motorised pendulum with a cubic torsional spring</em>; We introduce the <strong>Nonholonomic Locomotion - NNM (NL-NNM)</strong>: extract a <em>high-speed spectral seed</em> – where chassis oscillations vanish and the pendulum is neutrally stable – refine the periodic orbit, and continue the resulting <em>2D invariant manifold</em> via pseudo-arclength across <em>three slow centre manifolds</em> (stable for positive speed, neutral at zero, unstable for negative) from non-isolated rectilinear equilibria; We demonstrate the first NL-NNM for a moving non-holonomic robot: internal orbits produce a <em>pendulum–chassis choreography</em> whose <em>energy-dependent frequency shifts</em> and <em>harmonic richness</em> exceed linear predictions. Via <em>geometric phase</em>, each orbit yields undulatory straight-line motion. A <em>dual-loop control simulation</em> confirms autonomous path tracking with only the pendulum; Extending to dissipative regimes via <em>non-linear resonant modes</em> offers a path to high-efficiency locomotion in aquatic, aerial, legged, soft-bodied, and other robots.</div></div>","PeriodicalId":49842,"journal":{"name":"Mechatronics","volume":"110 ","pages":"Article 103366"},"PeriodicalIF":3.1,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144322976","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Robotized Incremental Sheet Forming trajectory control using deep neural network for force/torque compensator and task-space error tracking controller","authors":"Xuan Dung To ,&nbsp;Jefferson Roman Blanco ,&nbsp;Sandra Zimmer-Chevret ,&nbsp;Ghinwa Ouaidat ,&nbsp;Thibaut Raharijaona ,&nbsp;Farid Noureddine ,&nbsp;Micky Rakotondrabe","doi":"10.1016/j.mechatronics.2025.103360","DOIUrl":"10.1016/j.mechatronics.2025.103360","url":null,"abstract":"<div><div>In Robotized Incremental Sheet Forming (ISF), achieving precise geometrical accuracy is a challenging task due to trajectory tool center point (TCP) position errors at the forming tool attached to the robot’s end-effector. These errors primarily arise from external disturbance forces and torques generated during the interaction between the forming tool and the elastic metal sheet. While joint-torque space controllers can mitigate reaction forces and torques through dynamic modeling, joint-space control has inherent limitations, particularly for industrial high-load robots like the ABB IRB 8700. To overcome these challenges, this work implements an external force/torque (F/T) compensator in task-space using a deep neural network. The network predicts trajectory errors induced by reaction forces and torques measured via a 6-axis F/T sensor. Additionally, the forming tool’s trajectory is precisely monitored using a laser tracker, which serves as a feedback mechanism in a closed-loop task-space error-tracking controller. This controller detects and corrects trajectory deviations in real time. By integrating the F/T compensator and the task-space error-tracking controller, the proposed approach effectively compensates for reaction forces and torques while addressing additional errors introduced by other process-related factors. This integration results in significantly enhanced accuracy in robotic incremental forming processes.</div></div>","PeriodicalId":49842,"journal":{"name":"Mechatronics","volume":"110 ","pages":"Article 103360"},"PeriodicalIF":3.1,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144322975","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Fast switching and dynamic characteristics preservation of water hydraulic high-speed on-off valve using pressure-adaptive multistage voltage and sliding mode control","authors":"Xing Yang,&nbsp;Boyang Zhang,&nbsp;Defa Wu,&nbsp;Yinshui Liu","doi":"10.1016/j.mechatronics.2025.103384","DOIUrl":"10.1016/j.mechatronics.2025.103384","url":null,"abstract":"<div><div>Water hydraulic high-speed on-off valves (WHSVs) are crucial for managing fluid flow in water hydraulic manipulator systems. As the ambient pressure changes, the dynamic characteristics of WHSVs are affected, which reduces the overall control accuracy of the manipulator. To simultaneously achieve rapid switching and maintain consistent dynamic behavior of WHSVs under variable ambient pressure, a pressure-adaptive multistage voltage and sliding mode control (PMVS) algorithm is proposed. A sliding mode controller is utilized to precisely regulate the coil current at the pre-opening and holding current levels, significantly shortening the switching time of the WHSV. By optimizing the controller’s operation time and modifying the duty cycle of the excitation voltage, the switching time remains stable across different pressures. Based on the structure of the designed WHSV group, an innovative method combining a pressure sensor and a vibration sensor is proposed to capture the dynamic characteristics of the WHSV. Experimental validation demonstrates that the PMVS method efficiently controls the switching delay and regulates the excitation voltage. Dynamic characteristic tests of WHSVs under different pressures are conducted. The results show that PMVS effectively reduces the switching time of WHSVs. Comparative tests reveal that WHSVs driven by PMVS achieve an 86.3 % reduction in opening time and an 87.5 % reduction in closing time compared to conventional pulse width modulation (CPWM). Furthermore, PMVS ensures consistent dynamic characteristics within an ambient pressure range of 0 to 20 MPa, with an opening time deviation of 7.94 % and a closing time deviation of 3.03 %. The PMVS algorithm enables the WHSV to rapidly switch and preserve dynamic characteristics under variable ambient pressures.</div></div>","PeriodicalId":49842,"journal":{"name":"Mechatronics","volume":"110 ","pages":"Article 103384"},"PeriodicalIF":3.1,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144563671","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}

{"title":"Learning an inverse thermodynamic model for Pneumatic Artificial Muscles control","authors":"G. Wang ,&nbsp;R. Chalard ,&nbsp;J.A. Cifuentes ,&nbsp;M.T. Pham","doi":"10.1016/j.mechatronics.2025.103359","DOIUrl":"10.1016/j.mechatronics.2025.103359","url":null,"abstract":"<div><div>Pneumatic Artificial Muscles (PAMs) are highly nonlinear actuators widely used in robotics, rehabilitation, and other dynamic applications. Their complex behavior poses significant challenges for traditional system identification methods. Although machine learning techniques have shown remarkable success in modeling nonlinear systems, their black-box nature often leads to interpretability issues and susceptibility to overfitting. This study proposes a novel hybrid modeling approach that combines the strengths of analytical models with neural networks to capture the inverse thermodynamic behavior of PAMs. The results demonstrate that the hybrid model outperformed both analytical and purely neural network models. The obtained models were further used for model-based control design and the results show that the application of hybrid model improved the tracking performance.</div></div>","PeriodicalId":49842,"journal":{"name":"Mechatronics","volume":"110 ","pages":"Article 103359"},"PeriodicalIF":3.1,"publicationDate":"2025-10-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"144313090","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":3,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}