Pub Date : 2026-01-21DOI: 10.1016/j.conengprac.2026.106761
Shaoxun Liu , Shiyu Zhou , Mohamed Abdullah , Zhengsheng Liu , Hui Zhang , Rongrong Wang
The reliability of actuators and sensors is critical for the safe operation of heavy-legged robots (HLRs), yet few schemes handle concurrent actuator and sensor faults under model uncertainty. This study presents an integrated fault-tolerant control architecture that incorporates permanent magnet synchronous motor (PMSM) phase currents into the trajectory tracking loop. A complementary filter-based observer is developed to estimate HLR velocities and model residuals, while a fault gain loss estimator (FGLE) quantifies servo fault severity from PMSM current-controller residuals. An embedded radial basis function network further determines whether the HLR retains its fault-tolerant capability. Leveraging the estimated servo status, a field-oriented fault-tolerant controller is formulated to (1) seamlessly coordinate PMSM and HLR control, (2) sustain precise trajectory tracking under actuator and sensor failures, and (3) execute a controlled stop when driving capability is lost, thereby mitigating economic loss. Experimental validation on an electric-cylinder-driven HLR demonstrates that the proposed framework reduces RMSE and MAE by 11.7% and 10.2%, respectively, compared with conventional integrated frameworks, while ensuring safe shutdown without secondary damage.
{"title":"Evaluation of servo fault status and fault-tolerant control for heavy-legged robots under concurrent velocity sensor failures","authors":"Shaoxun Liu , Shiyu Zhou , Mohamed Abdullah , Zhengsheng Liu , Hui Zhang , Rongrong Wang","doi":"10.1016/j.conengprac.2026.106761","DOIUrl":"10.1016/j.conengprac.2026.106761","url":null,"abstract":"<div><div>The reliability of actuators and sensors is critical for the safe operation of heavy-legged robots (HLRs), yet few schemes handle concurrent actuator and sensor faults under model uncertainty. This study presents an integrated fault-tolerant control architecture that incorporates permanent magnet synchronous motor (PMSM) phase currents into the trajectory tracking loop. A complementary filter-based observer is developed to estimate HLR velocities and model residuals, while a fault gain loss estimator (FGLE) quantifies servo fault severity from PMSM current-controller residuals. An embedded radial basis function network further determines whether the HLR retains its fault-tolerant capability. Leveraging the estimated servo status, a field-oriented fault-tolerant controller is formulated to (1) seamlessly coordinate PMSM and HLR control, (2) sustain precise trajectory tracking under actuator and sensor failures, and (3) execute a controlled stop when driving capability is lost, thereby mitigating economic loss. Experimental validation on an electric-cylinder-driven HLR demonstrates that the proposed framework reduces RMSE and MAE by 11.7% and 10.2%, respectively, compared with conventional integrated frameworks, while ensuring safe shutdown without secondary damage.</div></div>","PeriodicalId":50615,"journal":{"name":"Control Engineering Practice","volume":"169 ","pages":"Article 106761"},"PeriodicalIF":4.6,"publicationDate":"2026-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146038364","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-21DOI: 10.1016/j.conengprac.2026.106785
Songmiao Li , Yangfan Zhou , Pengze Liu , Dan Ye , Bi Zhang , Xingang Zhao
Functional electrical stimulation (FES) has shown promise in restoring motor functions for patients with spinal cord injury and stroke. However, its clinical application is limited by insufficient accuracy in modeling muscle dynamics and the lack of robust control strategies under complex disturbances. To address these challenges, this study proposes a closed-loop framework that integrates high-precision modeling with strong robustness. A Hammerstein model enhanced by Kolmogorov-Arnold Networks (KAN) is constructed, where the explicit mathematical representation of KAN significantly improves the nonlinear dynamic modeling of muscle behavior. Additionally, a forgetting factor recursive least squares (FFRLS) algorithm is employed for online identification of time-varying parameters, achieving improved performance over traditional approaches. Further, a sliding-mode tube model predictive control (SMC-Tube MPC) strategy driven by surface electromyography (sEMG) feedback is developed. By combining the disturbance rejection capability of sliding mode control with the state constraint handling features of Tube-MPC, the proposed controller enables stable torque tracking under complex perturbations. The framework is validated on an experimental platform integrating a dynamometer, sEMG acquisition device, and electrical stimulator. Experiments with healthy subjects demonstrate high accuracy and strong robustness of the proposed system.
{"title":"KAN-Hammerstein model and tube-based model predictive control for robust torque tracking with sEMG feedback in an FES-assisted rehabilitation system","authors":"Songmiao Li , Yangfan Zhou , Pengze Liu , Dan Ye , Bi Zhang , Xingang Zhao","doi":"10.1016/j.conengprac.2026.106785","DOIUrl":"10.1016/j.conengprac.2026.106785","url":null,"abstract":"<div><div>Functional electrical stimulation (FES) has shown promise in restoring motor functions for patients with spinal cord injury and stroke. However, its clinical application is limited by insufficient accuracy in modeling muscle dynamics and the lack of robust control strategies under complex disturbances. To address these challenges, this study proposes a closed-loop framework that integrates high-precision modeling with strong robustness. A Hammerstein model enhanced by Kolmogorov-Arnold Networks (KAN) is constructed, where the explicit mathematical representation of KAN significantly improves the nonlinear dynamic modeling of muscle behavior. Additionally, a forgetting factor recursive least squares (FFRLS) algorithm is employed for online identification of time-varying parameters, achieving improved performance over traditional approaches. Further, a sliding-mode tube model predictive control (SMC-Tube MPC) strategy driven by surface electromyography (sEMG) feedback is developed. By combining the disturbance rejection capability of sliding mode control with the state constraint handling features of Tube-MPC, the proposed controller enables stable torque tracking under complex perturbations. The framework is validated on an experimental platform integrating a dynamometer, sEMG acquisition device, and electrical stimulator. Experiments with healthy subjects demonstrate high accuracy and strong robustness of the proposed system.</div></div>","PeriodicalId":50615,"journal":{"name":"Control Engineering Practice","volume":"169 ","pages":"Article 106785"},"PeriodicalIF":4.6,"publicationDate":"2026-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146038457","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-21DOI: 10.1016/j.conengprac.2026.106787
Bart van Laatum , Salim Msaad , Eldert J. van Henten , Robert D. Mcallister , Sjoerd Boersma
Uncertainty, if not explicitly accounted for in controller design, can significantly degrade the optimal control performance of greenhouse production systems. Scenario-based stochastic MPC (SMPC) addresses uncertainty by approximating its underlying probability distributions through sampling. However, SMPC rapidly becomes computationally intractable and can suffer from growing uncertainty with longer prediction horizons. Terminal costs and constraints ensure closed-loop performance of SMPC, but designing these for greenhouse systems is challenging since they rely on steady-state targets that often do not exist in greenhouse production systems. To overcome these challenges, this work introduces RL-SMPC, which uses reinforcement learning (RL) to learn a control policy that constructs both terminal region constraints and a terminal cost function. Additionally, this policy serves as a nonlinear feedback policy to attenuate uncertainty growth in the open-loop solution of scenario-based SMPC. RL-SMPC’s closed-loop performance is compared against standalone RL, MPC, and scenario-based SMPC on a greenhouse lettuce model under parametric uncertainty. Simulation results showed that RL-SMPC outperformed MPC across all prediction horizons and surpassed SMPC for horizons shorter than five hours. Moreover, the results indicated that at equal online computational cost, RL-SMPC outperformed SMPC.
{"title":"Stochastic model predictive control with reinforcement learning for greenhouse production systems under parametric uncertainty","authors":"Bart van Laatum , Salim Msaad , Eldert J. van Henten , Robert D. Mcallister , Sjoerd Boersma","doi":"10.1016/j.conengprac.2026.106787","DOIUrl":"10.1016/j.conengprac.2026.106787","url":null,"abstract":"<div><div>Uncertainty, if not explicitly accounted for in controller design, can significantly degrade the optimal control performance of greenhouse production systems. Scenario-based stochastic MPC (SMPC) addresses uncertainty by approximating its underlying probability distributions through sampling. However, SMPC rapidly becomes computationally intractable and can suffer from growing uncertainty with longer prediction horizons. Terminal costs and constraints ensure closed-loop performance of SMPC, but designing these for greenhouse systems is challenging since they rely on steady-state targets that often do not exist in greenhouse production systems. To overcome these challenges, this work introduces RL-SMPC, which uses reinforcement learning (RL) to learn a control policy that constructs both terminal region constraints and a terminal cost function. Additionally, this policy serves as a nonlinear feedback policy to attenuate uncertainty growth in the open-loop solution of scenario-based SMPC. RL-SMPC’s closed-loop performance is compared against standalone RL, MPC, and scenario-based SMPC on a greenhouse lettuce model under parametric uncertainty. Simulation results showed that RL-SMPC outperformed MPC across all prediction horizons and surpassed SMPC for horizons shorter than five hours. Moreover, the results indicated that at equal online computational cost, RL-SMPC outperformed SMPC.</div></div>","PeriodicalId":50615,"journal":{"name":"Control Engineering Practice","volume":"169 ","pages":"Article 106787"},"PeriodicalIF":4.6,"publicationDate":"2026-01-21","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146038362","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-20DOI: 10.1016/j.conengprac.2026.106769
S.K. Mallipeddi , M. Menghini , S. Simani , P. Castaldi
Path following for underwater vehicles remains a significant challenge due to underactuation in the sway and heave directions. Most existing approaches rely on line-of-sight guidance to address this issue. In this paper, we explore an alternative approach using kinematic guidance, based on virtual reference point guidance, wherein a fictitious point offset from the vehicle’s center of rotation is used to reformulate the kinematic control problem and mitigate underactuation constraints. While this concept has been explored to some extent, previous works have largely overlooked the impact of the vehicle’s attitude. To address this limitation, we propose a solution that simultaneously accounts for the vehicle’s attitude while minimizing cross-track error by defining the error dynamics in the body reference frame, which enables direct control of yaw and sway through yaw rate actuation. A model predictive controller is designed to optimize both attitude stabilization and trajectory tracking performance and is enhanced with an adaptive extended Kalman filter-like observer to estimate the sideslip caused by sea currents and external disturbances. The proposed controller is evaluated under the influence of sea currents and modeling uncertainties, and compared to an existing method from the literature, demonstrating its effectiveness in maintaining path-following accuracy while stabilizing the attitude in the presences of the sea currents.
{"title":"Kinematic guidance using virtual reference point for underactuated marine vehicles with sideslip compensation","authors":"S.K. Mallipeddi , M. Menghini , S. Simani , P. Castaldi","doi":"10.1016/j.conengprac.2026.106769","DOIUrl":"10.1016/j.conengprac.2026.106769","url":null,"abstract":"<div><div>Path following for underwater vehicles remains a significant challenge due to underactuation in the sway and heave directions. Most existing approaches rely on line-of-sight guidance to address this issue. In this paper, we explore an alternative approach using kinematic guidance, based on virtual reference point guidance, wherein a fictitious point offset from the vehicle’s center of rotation is used to reformulate the kinematic control problem and mitigate underactuation constraints. While this concept has been explored to some extent, previous works have largely overlooked the impact of the vehicle’s attitude. To address this limitation, we propose a solution that simultaneously accounts for the vehicle’s attitude while minimizing cross-track error by defining the error dynamics in the body reference frame, which enables direct control of yaw and sway through yaw rate actuation. A model predictive controller is designed to optimize both attitude stabilization and trajectory tracking performance and is enhanced with an adaptive extended Kalman filter-like observer to estimate the sideslip caused by sea currents and external disturbances. The proposed controller is evaluated under the influence of sea currents and modeling uncertainties, and compared to an existing method from the literature, demonstrating its effectiveness in maintaining path-following accuracy while stabilizing the attitude in the presences of the sea currents.</div></div>","PeriodicalId":50615,"journal":{"name":"Control Engineering Practice","volume":"169 ","pages":"Article 106769"},"PeriodicalIF":4.6,"publicationDate":"2026-01-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146038455","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-20DOI: 10.1016/j.conengprac.2025.106724
Gustavo G. Koch , Lucas Borin , Caio Osório , Mokthar Aly , Margarita Norambuena , Jose Rodriguez , Fernanda Carnieluti , Humberto Pinheiro , Ricardo C.L.F. Oliveira , Vinícius F. Montagner
This paper introduces a new methodology for designing robust current controllers for grid-connected converters (GCCs) with LCL filters, ensuring suitable operation from strong to very weak grid conditions. The approach combines i) a polytopic plant model accounting for control delay and parametric uncertainties, ii) improved linear matrix inequality (LMI) synthesis conditions for robust pole placement, and iii) experimental validation via Controller Hardware-in-the-Loop (C-HIL). The LMI-based design integrated with C-HIL guarantees theoretical robustness and provides practical insights on the performance of the controller with unmodeled dynamics and nonlinearities, enhancing the robustness-performance trade-off while reducing costs and risks. Experimental results for a GCC under a grid with large impedance uncertainty and voltage harmonics show that traditional LMI techniques produce higher control gains causing persistent saturation of the actuator and degrading the performance in real implementation. Conversely, the proposed methodology ensures compliance with reference tracking, harmonics rejection, and voltage dip recovery, even under very weak grids (short-circuit ratio (SCR) = 1). Compared to methods relying on LMIs and C-HIL, the proposal is much superior, computing control gains at least 20 times faster through a fully deterministic convex optimization, while ensuring high-performance when implemented online on off-the-shelf digital signal processors.
{"title":"Improved control of grid-connected converters from strong to very weak conditions integrating more effective LMIs and C-HIL","authors":"Gustavo G. Koch , Lucas Borin , Caio Osório , Mokthar Aly , Margarita Norambuena , Jose Rodriguez , Fernanda Carnieluti , Humberto Pinheiro , Ricardo C.L.F. Oliveira , Vinícius F. Montagner","doi":"10.1016/j.conengprac.2025.106724","DOIUrl":"10.1016/j.conengprac.2025.106724","url":null,"abstract":"<div><div>This paper introduces a new methodology for designing robust current controllers for grid-connected converters (GCCs) with LCL filters, ensuring suitable operation from strong to very weak grid conditions. The approach combines i) a polytopic plant model accounting for control delay and parametric uncertainties, ii) improved linear matrix inequality (LMI) synthesis conditions for robust pole placement, and iii) experimental validation via Controller Hardware-in-the-Loop (C-HIL). The LMI-based design integrated with C-HIL guarantees theoretical robustness and provides practical insights on the performance of the controller with unmodeled dynamics and nonlinearities, enhancing the robustness-performance trade-off while reducing costs and risks. Experimental results for a GCC under a grid with large impedance uncertainty and voltage harmonics show that traditional LMI techniques produce higher control gains causing persistent saturation of the actuator and degrading the performance in real implementation. Conversely, the proposed methodology ensures compliance with reference tracking, harmonics rejection, and voltage dip recovery, even under very weak grids (short-circuit ratio (SCR) = 1). Compared to methods relying on LMIs and C-HIL, the proposal is much superior, computing control gains at least 20 times faster through a fully deterministic convex optimization, while ensuring high-performance when implemented online on off-the-shelf digital signal processors.</div></div>","PeriodicalId":50615,"journal":{"name":"Control Engineering Practice","volume":"169 ","pages":"Article 106724"},"PeriodicalIF":4.6,"publicationDate":"2026-01-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146038454","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-19DOI: 10.1016/j.conengprac.2026.106792
Gengchen Liu , Song Gao , Junheng Jiang , Zhangmin Luo , Gang Jiang
Accurate path planning is particularly important for unmanned vehicles in complex mountainous environments. Compared with two-dimensional terrain, mountainous three-dimensional terrain not only introduces more uncertainty but also interference from dynamic obstacles, which dramatically increases the difficulty of path planning. As such, conventional planning methods often struggle to identify efficient solutions. Although path planning techniques utilizing deep reinforcement learning have provided new strategies for solving such problems, existing algorithms face a variety of challenges, including poor network stability, susceptibility to gradient explosion, insufficient reward guidance, and an imbalance between exploration and utilization. To overcome these issues, this paper introduces three novel contributions. First, the dueling double DQN is structurally optimized, and various techniques are introduced to prevent instability and gradient explosion. Second, a new reward function is developed to combine the Bessel hierarchical A* path guidance algorithm with the artificial potential field method, enabling unmanned vehicles to identify the optimal path while dynamically avoiding obstacles. Finally, a chaotic annealing multi-phased strategy is proposed as an action selection policy, which gradually transitions from the exploration stage to the exploitation stage by optimizing the balance between the two as the learning process advances. In addition, a 3D terrain model based on a real mountain environment was generated using the grayscale map algorithm. A series of simulation experiments were conducted to evaluate the performance of the proposed method, as measured by search efficiency, success rate, and path quality. A comparative analysis and comparison with existing DRL path planning algorithms was also performed to provide additional insights.
{"title":"Mountain UGV path planning via optimized dueling double DQN (D3QN): Structural optimization, path-guided rewards, and phased action policy","authors":"Gengchen Liu , Song Gao , Junheng Jiang , Zhangmin Luo , Gang Jiang","doi":"10.1016/j.conengprac.2026.106792","DOIUrl":"10.1016/j.conengprac.2026.106792","url":null,"abstract":"<div><div>Accurate path planning is particularly important for unmanned vehicles in complex mountainous environments. Compared with two-dimensional terrain, mountainous three-dimensional terrain not only introduces more uncertainty but also interference from dynamic obstacles, which dramatically increases the difficulty of path planning. As such, conventional planning methods often struggle to identify efficient solutions. Although path planning techniques utilizing deep reinforcement learning have provided new strategies for solving such problems, existing algorithms face a variety of challenges, including poor network stability, susceptibility to gradient explosion, insufficient reward guidance, and an imbalance between exploration and utilization. To overcome these issues, this paper introduces three novel contributions. First, the dueling double DQN is structurally optimized, and various techniques are introduced to prevent instability and gradient explosion. Second, a new reward function is developed to combine the Bessel hierarchical A* path guidance algorithm with the artificial potential field method, enabling unmanned vehicles to identify the optimal path while dynamically avoiding obstacles. Finally, a chaotic annealing multi-phased strategy is proposed as an action selection policy, which gradually transitions from the exploration stage to the exploitation stage by optimizing the balance between the two as the learning process advances. In addition, a 3D terrain model based on a real mountain environment was generated using the grayscale map algorithm. A series of simulation experiments were conducted to evaluate the performance of the proposed method, as measured by search efficiency, success rate, and path quality. A comparative analysis and comparison with existing DRL path planning algorithms was also performed to provide additional insights.</div></div>","PeriodicalId":50615,"journal":{"name":"Control Engineering Practice","volume":"169 ","pages":"Article 106792"},"PeriodicalIF":4.6,"publicationDate":"2026-01-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146038360","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
With the advancements towards automated and autonomous driving system, this paper develops a novel steer-by-wire (SbW) configuration based on a dual three-phase permanent magnet synchronous motor (DTP-PMSM). This system incorporates an innovative triple redundant orthogonal decoupling technology. The DTP-PMSM is decoupled into three independently controllable two-phase orthogonal motors that drive the steering system through rigid coaxial output. To improve the steering angle tracking accuracy and anti-interference capability of this triple redundant SbW system, this paper propose a two-layer control strategy. The outer layer features an Angle Tracking Controller (ATC) utilizing a non-singular fast terminal sliding mode approach combined with an extended state observer. The ATC tracks the steering angle and outputs the target current. The inner layer employs a Torque Synchronous Controller (TSC), which allocates the target current as reference torque signals to the three redundant motors. Taking into account the delay of the signal, this paper introduce an improved generalized predictive torque synchronization algorithm with mean deviation coupling, optimized via a wavelet neural network. This algorithm balances the output torque between the three redundant motors, suppresses torque asynchrony caused by parameter variations, disturbances, and faults, and improves steering tracking performance. Crucially, to prevent imbalance resulting from the fixed gain in the deviation-coupling structure, this paper propose a wavelet neural network compensator. This compensator dynamically optimizes the structural gain, enabling rapid and precise deviation compensation to achieve fast elimination of torque errors between the three redundant motors. Experimental results demonstrate that the triple redundant motor system achieves rapid torque synchronization and significantly improves the steering angle tracking performance of the SbW system.
{"title":"Redundant torque syncronization and steering angle tracking strategy for dual three phase steer-by-wire system","authors":"Haoyu Sun, Wanzhong Zhao, Chunyan Wang, Zhongkai Luan, Weihe Liang, Ziyu Zhang, Xiaochuan Zhou, Yukai Chu","doi":"10.1016/j.conengprac.2026.106784","DOIUrl":"10.1016/j.conengprac.2026.106784","url":null,"abstract":"<div><div>With the advancements towards automated and autonomous driving system, this paper develops a novel steer-by-wire (SbW) configuration based on a dual three-phase permanent magnet synchronous motor (DTP-PMSM). This system incorporates an innovative triple redundant orthogonal decoupling technology. The DTP-PMSM is decoupled into three independently controllable two-phase orthogonal motors that drive the steering system through rigid coaxial output. To improve the steering angle tracking accuracy and anti-interference capability of this triple redundant SbW system, this paper propose a two-layer control strategy. The outer layer features an Angle Tracking Controller (ATC) utilizing a non-singular fast terminal sliding mode approach combined with an extended state observer. The ATC tracks the steering angle and outputs the target current. The inner layer employs a Torque Synchronous Controller (TSC), which allocates the target current as reference torque signals to the three redundant motors. Taking into account the delay of the signal, this paper introduce an improved generalized predictive torque synchronization algorithm with mean deviation coupling, optimized via a wavelet neural network. This algorithm balances the output torque between the three redundant motors, suppresses torque asynchrony caused by parameter variations, disturbances, and faults, and improves steering tracking performance. Crucially, to prevent imbalance resulting from the fixed gain in the deviation-coupling structure, this paper propose a wavelet neural network compensator. This compensator dynamically optimizes the structural gain, enabling rapid and precise deviation compensation to achieve fast elimination of torque errors between the three redundant motors. Experimental results demonstrate that the triple redundant motor system achieves rapid torque synchronization and significantly improves the steering angle tracking performance of the SbW system.</div></div>","PeriodicalId":50615,"journal":{"name":"Control Engineering Practice","volume":"169 ","pages":"Article 106784"},"PeriodicalIF":4.6,"publicationDate":"2026-01-19","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146038453","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-18DOI: 10.1016/j.conengprac.2025.106752
Kinza Nazir , Yong-Woon Kim , Chul-Ung Kang , Yung-Cheol Byun
Automated Guided Vehicles (AGVs) are increasingly deployed in modern industrial environments, where precise and adaptive motion control is critical for efficient operation. This study presents a data-driven framework for intelligent tuning of Proportional-Integral (PI) controllers using a voting-based ensemble of machine learning models. The proposed framework leverages experimental AGV data, augmented using Gaussian jittering, to train an ensemble regressor comprising k-Nearest Neighbors, Random Forest, and Support Vector Regressor. Feature engineering techniques and SHAP-based interpretability were applied to ensure robust performance and transparency. Offline and real-world experiments were conducted across multiple speed levels to validate the model’s accuracy and generalization. In offline experiments at 25 m per minute, the model achieved an R2 score of 0.8635, RMSE of 0.1625, MAE of 0.1124, and MSE of 0.0264. Results show that the ensemble model significantly outperforms traditional tuning methods, with substantial improvements in error reduction and predictive reliability. Real-world testing further confirmed the framework’s effectiveness, where iterative deployment across multiple tracks resulted in progressively decreasing prediction errors and successful identification of optimal control parameters. This framework offers a scalable, low-latency solution for PI control in dynamic industrial applications.
自动导引车（agv）越来越多地应用于现代工业环境中，其中精确和自适应运动控制对于高效运行至关重要。本研究提出了一个数据驱动的框架，用于使用基于投票的机器学习模型集成来智能调整比例积分（PI）控制器。所提出的框架利用实验AGV数据，使用高斯抖动增强，训练由k-近邻、随机森林和支持向量回归器组成的集成回归器。应用特征工程技术和基于shap的可解释性来确保稳健的性能和透明度。为了验证模型的准确性和泛化性，在多个速度水平下进行了离线和现实世界的实验。在25 m / min的离线实验中，模型的R2得分为0.8635，RMSE为0.1625，MAE为0.1124，MSE为0.0264。结果表明，该集成模型明显优于传统的调谐方法，在减少误差和预测可靠性方面有显著提高。实际测试进一步证实了该框架的有效性，其中跨多个轨道的迭代部署导致预测误差逐渐减少，并成功识别出最优控制参数。该框架为动态工业应用中的PI控制提供了可扩展的低延迟解决方案。
{"title":"Intelligent PI control for trajectory regulation in autonomous vehicles using a voting-based ensemble of statistical learning models","authors":"Kinza Nazir , Yong-Woon Kim , Chul-Ung Kang , Yung-Cheol Byun","doi":"10.1016/j.conengprac.2025.106752","DOIUrl":"10.1016/j.conengprac.2025.106752","url":null,"abstract":"<div><div>Automated Guided Vehicles (AGVs) are increasingly deployed in modern industrial environments, where precise and adaptive motion control is critical for efficient operation. This study presents a data-driven framework for intelligent tuning of Proportional-Integral (PI) controllers using a voting-based ensemble of machine learning models. The proposed framework leverages experimental AGV data, augmented using Gaussian jittering, to train an ensemble regressor comprising k-Nearest Neighbors, Random Forest, and Support Vector Regressor. Feature engineering techniques and SHAP-based interpretability were applied to ensure robust performance and transparency. Offline and real-world experiments were conducted across multiple speed levels to validate the model’s accuracy and generalization. In offline experiments at 25 m per minute, the model achieved an R<sup>2</sup> score of 0.8635, RMSE of 0.1625, MAE of 0.1124, and MSE of 0.0264. Results show that the ensemble model significantly outperforms traditional tuning methods, with substantial improvements in error reduction and predictive reliability. Real-world testing further confirmed the framework’s effectiveness, where iterative deployment across multiple tracks resulted in progressively decreasing prediction errors and successful identification of optimal control parameters. This framework offers a scalable, low-latency solution for PI control in dynamic industrial applications.</div></div>","PeriodicalId":50615,"journal":{"name":"Control Engineering Practice","volume":"169 ","pages":"Article 106752"},"PeriodicalIF":4.6,"publicationDate":"2026-01-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146038452","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-18DOI: 10.1016/j.conengprac.2026.106765
Jian Li , Defu Kong , Zezhong Han , Linda Shen , Xinying Xue , Debin Chen , Zhiwei Zheng , Shasha Luo , Yunze Tan , Riwei Ye , Shuwan Cui , Yuanzhao Chen , Zhenfang Mao , Yang Yang , Chengzhang Su , Peng Liang
To address the requirement for coordinated hip-knee motion in lower limb rehabilitation training for bedridden stroke patients, this paper develops a horizontal lower limb rehabilitation robot based on a symmetrical five-bar slider mechanism. The research focuses on its coordinated motion control and human-machine interaction compliance. By establishing a human-machine coupled dynamic model, the interaction torque transmission mechanism was quantified. Based on this, a hierarchical control framework was constructed: during passive training, a genetic algorithm-optimized PID was employed to achieve preset trajectory tracking; during active training, a second-order impedance control strategy incorporating weighted fusion of pressure/angle information was proposed to adjust interaction forces. Multiple passive tests yielded average relative errors between theoretical calculations and measured interaction torques of 3.0% for the hip joint and 1.5% for the knee joint. Trajectory errors were reduced by 19.7% for the knee joint in hip-knee flexion mode, 36% in horizontal gait mode, and 6.3% for the thigh in bridge active training, with a maximum deviation of 3.98∘. Follow-up assessments of patients undergoing rehabilitation training revealed significant improvements in ASIA scores following robot-assisted training. This validated the accuracy of the proposed dynamic model and the effectiveness and robustness of the hierarchical control strategy, demonstrating the system’s capability to provide reliable lower-limb rehabilitation assistance for bedridden patients.
{"title":"Hip-knee coordination control and clinical validation of a horizontal lower-limb rehabilitation robot based on human-machine coupling dynamics modeling","authors":"Jian Li , Defu Kong , Zezhong Han , Linda Shen , Xinying Xue , Debin Chen , Zhiwei Zheng , Shasha Luo , Yunze Tan , Riwei Ye , Shuwan Cui , Yuanzhao Chen , Zhenfang Mao , Yang Yang , Chengzhang Su , Peng Liang","doi":"10.1016/j.conengprac.2026.106765","DOIUrl":"10.1016/j.conengprac.2026.106765","url":null,"abstract":"<div><div>To address the requirement for coordinated hip-knee motion in lower limb rehabilitation training for bedridden stroke patients, this paper develops a horizontal lower limb rehabilitation robot based on a symmetrical five-bar slider mechanism. The research focuses on its coordinated motion control and human-machine interaction compliance. By establishing a human-machine coupled dynamic model, the interaction torque transmission mechanism was quantified. Based on this, a hierarchical control framework was constructed: during passive training, a genetic algorithm-optimized PID was employed to achieve preset trajectory tracking; during active training, a second-order impedance control strategy incorporating weighted fusion of pressure/angle information was proposed to adjust interaction forces. Multiple passive tests yielded average relative errors between theoretical calculations and measured interaction torques of 3.0% for the hip joint and 1.5% for the knee joint. Trajectory errors were reduced by 19.7% for the knee joint in hip-knee flexion mode, 36% in horizontal gait mode, and 6.3% for the thigh in bridge active training, with a maximum deviation of 3.98<sup>∘</sup>. Follow-up assessments of patients undergoing rehabilitation training revealed significant improvements in ASIA scores following robot-assisted training. This validated the accuracy of the proposed dynamic model and the effectiveness and robustness of the hierarchical control strategy, demonstrating the system’s capability to provide reliable lower-limb rehabilitation assistance for bedridden patients.</div></div>","PeriodicalId":50615,"journal":{"name":"Control Engineering Practice","volume":"169 ","pages":"Article 106765"},"PeriodicalIF":4.6,"publicationDate":"2026-01-18","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"146038456","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2026-01-17DOI: 10.1016/j.conengprac.2026.106791
Hui Xie , Sihan Chen , Gang Shen , Shuhui Fei , Yu Tang , Yongcun Guo , Yuanjing He
To address the challenges of inconsistent shoe-approaching motion and excessive transient impacts in the operation of multi-channel braking systems (MCBS) for mine hoists, this investigation presents a novel hybrid shoe-approaching/pressure control strategy utilizing real-time braking pressure feedback. First, the braking process is divided into two distinct stages: shoe-approaching motion and contact compression. A hybrid position/force switching control scheme, relying on braking pressure feedback, is developed using hysteresis switching principle. Second, an online fastest shoe-approaching trajectory planning algorithm is designed with a nonlinear filter, and a three-loop shoe-approaching control strategy is proposed, which consists of an outer loop for shoe-approaching trajectory planning, an inner loop for position tracking of brake, and a cross coupled loop for multi-channel synchronous shoe-approaching motion. Finally, two sets of comparative experiments are carried out on the multi-channel braking test bench of the hoist. The experimental results indicate that, compared with the traditional braking control mode, the proposed braking control strategy can effectively suppress the braking transient impact, shorten the shoe-approaching time, and enhance the consistency of shoe-approaching motion of the MCBS.
{"title":"Hybrid shoe-approaching and pressure control strategy for multi-channel braking systems of mine hoists","authors":"Hui Xie , Sihan Chen , Gang Shen , Shuhui Fei , Yu Tang , Yongcun Guo , Yuanjing He","doi":"10.1016/j.conengprac.2026.106791","DOIUrl":"10.1016/j.conengprac.2026.106791","url":null,"abstract":"<div><div>To address the challenges of inconsistent shoe-approaching motion and excessive transient impacts in the operation of multi-channel braking systems (MCBS) for mine hoists, this investigation presents a novel hybrid shoe-approaching/pressure control strategy utilizing real-time braking pressure feedback. First, the braking process is divided into two distinct stages: shoe-approaching motion and contact compression. A hybrid position/force switching control scheme, relying on braking pressure feedback, is developed using hysteresis switching principle. Second, an online fastest shoe-approaching trajectory planning algorithm is designed with a nonlinear filter, and a three-loop shoe-approaching control strategy is proposed, which consists of an outer loop for shoe-approaching trajectory planning, an inner loop for position tracking of brake, and a cross coupled loop for multi-channel synchronous shoe-approaching motion. Finally, two sets of comparative experiments are carried out on the multi-channel braking test bench of the hoist. The experimental results indicate that, compared with the traditional braking control mode, the proposed braking control strategy can effectively suppress the braking transient impact, shorten the shoe-approaching time, and enhance the consistency of shoe-approaching motion of the MCBS.</div></div>","PeriodicalId":50615,"journal":{"name":"Control Engineering Practice","volume":"169 ","pages":"Article 106791"},"PeriodicalIF":4.6,"publicationDate":"2026-01-17","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"145979892","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":2,"RegionCategory":"计算机科学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
扫码关注我们
求助内容：
应助结果提醒方式：
京公网安备 11010802042870号