Reliable pipeline inspection is critical to safe energy transportation, but is constrained by long distances, complex terrain, and risks to human inspectors. Unmanned aerial vehicles provide a flexible sensing platform, yet reliable autonomous inspection remains challenging. This paper presents an autonomous quadrotor near-proximity pipeline inspection framework for three-dimensional scenarios based on image-based visual servoing model predictive control (VMPC). A unified predictive model couples quadrotor dynamics with image feature kinematics, enabling direct image-space prediction within the control loop. To address low-rate visual updates, measurement noise, and environmental uncertainties, an extended-state Kalman filtering scheme with image feature prediction (ESKF-PRE) is developed, and the estimated lumped disturbances are incorporated into the VMPC prediction model, yielding the ESKF-PRE-VMPC framework. A terrain-adaptive velocity design is introduced to maintain the desired cruising speed while generating vertical velocity references over unknown terrain slopes without prior terrain information. The framework is validated in high-fidelity Gazebo simulations and real-world experiments. In real-world tests, the proposed method reduces RMSE by 52.63

This article considers the current control problem of switched reluctance motors (SRMs) with structure-repetitive uncertainties. A novel discrete-time iterative learning observer (ILO)-based current control approach is proposed. First, a discrete-time ILO is designed to estimate the uncertainties in the current dynamics. By leveraging the information of previous estimates, the ILO can accurately estimate uncertainties that exhibit structure-repetitive characteristics. Then, a discrete-time composite current controller is constructed, wherein the estimation provided by the ILO is incorporated to compensate for the uncertainty. Compared with existing disturbance observer-based control schemes, the proposed method fully exploits the structural repetitiveness of uncertainties and thus effectively enhances the current tracking performance of SRMs. Rigorous analysis proves ultimate boundedness of the current tracking error and its convergence to zero under vanishing disturbance variations. Simulations and experiments demonstrate significant improvements of the proposed method in SRM current tracking performance.

We propose PRED-MPPI, the first MPPI variant that seamlessly integrates real-time disturbance preview and adaptive discretization for quadrotor tracking control under significant model inaccuracies and time-varying disturbances. Unlike prior MPPI variants (e.g., mathcal{L}_1-MPPI, DA-MPPI), which assume constant or matched disturbances, PRED-MPPI leverages a high-order Generalized Extended State Observer for disturbance preview and a Variable Discretization Grid (VDG) to reduce computation and control variance. The synergy enables real-time (50 Hz) quadrotor control under time-varying and mismatched disturbances. Extensive comparative simulation and real-world Crazyflie experiments demonstrate substantial performance gains. In AirSim simulation, PRED-MPPI reduces computation time by over 30%, and mean RMSE by 10.3%, 13.5%, and 14.6% compared to baseline MPPI, and by 2.59%, 3.62%, and 5.80% compared to DA-MPPI across three representative scenarios. In real-world Crazyflie experiments, for ground-effect-disturbed hovering, PRED-MPPI reduces mean and standard deviation (Std) of X–Y plane error by 14.2%/17.9% and 6.03%/21.6% compared to MPPI and DA-MPPI; for fan-induced wind experiments, PRED-MPPI yields improvements of 23.4%/36.8% and 13.8%/25.0% in RMSE and tracking error Std. These results establish PRED-MPPI as the first disturbance-preview MPPI achieving real-world UAV robustness and efficiency, paving the way for deployment on resource-limited robotic platforms. GitHub page with videos is at https://pred-mppi.github.io/

Accurate trajectory tracking control of mobile robots is a complex task due to inherent system nonlinearities, modeling uncertainties, and external disturbances. To tackle these challenges, this paper proposes a neural network-based residual learning approach (NNRes) for high-fidelity and efficient system modeling. Unlike full model learning, the NNRes framework focuses on learning only the residual error between a nominal physical model and the actual system dynamics, thus preserving prior knowledge while enhancing adaptability. This hybrid model, combining the physical model and learned residual, is embedded into a Nonlinear Model Predictive Control (NMPC) scheme to improve control accuracy and robustness. The proposed method is systematically compared with state-of-the-art methods: full model learning using NNs (NNFull) and residual learning using Gaussian Processes (GPRes). Comparative simulation results demonstrate that NNRes achieves superior trajectory tracking control performance, enhanced robustness, and lower computational cost. These findings validate the effectiveness of NNRes within NMPC frameworks and offer valuable insights for the intelligent control of complex robotic systems.

Classical phase-locked loops (PLLs) employing proportional-integral (PI) loop filters achieve only asymptotic convergence, resulting in limited transient speed and theoretically infinite settling time. To address these limitations, this article proposes a nonsmooth loop filter for fast and accurate PLL operation. A rigorous Lyapunov-based analysis is conducted, showing finite-time stabilization under different position reference signal inputs. The proposed nonsmooth PLL is further validated through comparative simulations and experiments conducted on a 2 kW interior permanent magnet synchronous motor (IPMSM) platform, showing faster phase and frequency convergence (up to 27.5% improvement in recovery time under speed changes) and smaller estimation error (up to 32% improvement in position error under speed changes) compared with conventional PI-based PLLs. Owing to its concise structure and analytically verified stability, the proposed nonsmooth PLL is expected to provide a practical and scalable solution for high-performance industrial electronic systems.

Estimating canopy structure - leaf inclination distribution (LIDFa), leaf area index (LAI), and fractional vegetation cover (FCover) - is vital for breeding, yet the added value of multi-angular UAV sensing over nadir-only baselines remains insufficiently quantified. This study developed a UAV-based multi-angular inversion framework that derived high-resolution bidirectional reflectance factors (BRF) from oblique photogrammetry and fitted a kernel-driven BRDF model to characterize reflectance anisotropy. Using transfer learning across cultivars and dates, we compared the retrieval performance of multi-angle versus nadir-only baselines for LIDFa, LAI, and FCover. BRDF model simulations agreed well with airborne BRF (optimal R 2 > 0.80, RRMSE < 0.2 ) and exhibited anisotropy consistent with ground measurements. The comparative analysis demonstrated that multi-angle observations significantly improved retrieval accuracy for LAI (R 2 = 0.59 vs. 0.38 for the best MA and NAD models, respectively) and LIDFa (R 2 = 0.46 vs. 0.37). For FCover, both configurations achieved high accuracy (R 2 ≥ 0.73), with MA models providing marginal gains (R 2 = 0.75). Methodologically, CNN-based transfer learning proved most effective for LAI and FCover, while a Random Forest model using raw multi-angle spectra yielded the best results for LIDFa. Optimal viewing configurations were trait-dependent, generally favoring forward scattering directions with zenith angles between 15° and 45°. These results indicate that kernel-driven BRDF modeling effectively captures spectral anisotropy in dense wheat canopies, and that multi-angular observations provide a distinct advantage for retrieving structural parameters with complex scattering behaviors, such as LAI and LIDFa.

This paper develops a dynamic obstacle avoidance approach for mobile robots by combining monocular depth estimation with an improved Artificial Potential Field (APF). Depth maps inferred from single-view RGB images via the SOTA Depth Anything model are fused with YOLO-based object detection to achieve low-cost, calibration-free, real-time target perception, compared against stereo vision and LiDAR based methods. To address failure modes common in monocular navigation, such as ground-obstacle ambiguity and depth noise near humans, the proposed APF incorporates virtual-depth expansion, human-safety filtering, and a lateral repulsive mechanism that improves maneuverability in cluttered and previously unobserved environments. Integrated in a real-time ROS2 pipeline, the system demonstrates reliable human tracking and collision-free navigation in complex scenes, highlighting its practicality for safe operation in unstructured settings. Demonstration videos for this work are available at: https://drive.google.com/drive/folders/1pCFMlwMjSb3mTV3E7ItHfkAeiEmdYvlt?usp=sharing

Accurate and high-rate state estimation is crucial for quadrotors due to its central role in decision, planning and control, which is challenging due to unknown disturbances. Disturbances include internal ones caused by model uncertainties such as mass or lift-coefficient mismatch, and external ones caused by unknown wind or striking forces. To address these challenges, we propose a disturbance-resilient and computationally efficient state estimation method, termed fast and robust moving horizon estimation (FR-MHE), and validate it via onboard experiments on a Crazyflie quadrotor with its resource-constrained STM32F405 MCU. Disturbances are first augmented into quadrotor dynamics to enhance generalization and robustness. We then design a lightweight computational framework via a single-iteration gradient descent to overcome the low computational efficiency of MHE, which otherwise cannot meet the controller’s 100 Hz state estimation demand. Extensive experiments across ten scenarios, including hovering under internal disturbances, wind gusts, payload changes, and striking forces, and flying four typical trajectories under wind gusts, changed paddles, and damaged paddles, demonstrate that, while maintaining estimation performance of Robust MHE (RMHE), FR-MHE improves computational efficiency by 55-126 $\mathbf{\times}$ and reaches sub-millisecond, achieving a runtime comparable to robust extended Kalman filter but with better state estimation capability. Videos are at https://superace0916.github.io/.

Safe motion planning in cluttered environments under unknown disturbances remains a major challenge for robotic systems. Model predictive path integral (MPPI) control is attractive for non-convex planning, yet its penalty-based formulation cannot provide deterministic safety guarantees. This paper presents Disturbance-Aware Safe MPPI (DAS-MPPI), a robust sampling-based framework that couples MPPI with Hamilton–Jacobi reachability to enforce reachability-certified safety under unknown disturbances. Unlike prior methods that assume fixed worst-case disturbance bounds, DAS-MPPI formulates the safety game against a convergent disturbance-estimation error set. This dynamically shrinks the backward reachable tube (BRT), reducing conservatism and expanding the navigable free space. Furthermore, we embed this adaptive BRT into MPPI as both a rollout filter and a final control filter, while using online disturbance estimates to compensate for lumped disturbances and mitigate steady-state tracking errors. Extensive comparative simulations show that DAS-MPPI substantially reduces unnecessary safety interventions compared with state-of-the-art baselines. Under moderate and time-varying disturbances, it improves both planning efficiency and success rate, while under the prescribed worst-case disturbance bound it maintains robust collision-avoidance performance. Hardware experiments on a quadruped robot further demonstrate its practical applicability in real-world scenarios.

Reconstituted stem silk (ReS) is a transformative material in the tobacco industry, where impurity detection is critical for ensuring product quality. However, automating this task via deep learning faces three major challenges: complex image backgrounds, extremely small impurities, and the absence of annotated datasets. To overcome these issues, this paper introduces three novel modules: a Coordinate Attention Encoding module for capturing global context, a Multi-Scale Downsampling module for preserving fine details, and a Coordinate-Focus Spatial Pyramid Pooling module for position-aware, multiscale feature fusion. Furthermore, a dedicated annotated ReS dataset is constructed and released to support this research. Comprehensive experiments validate that the proposed method significantly outperforms state-of-the-art models, achieving an 11.8% improvement in mAP and demonstrating a strong potential for industrial inspection.

The sliding mode controller (SMC) is inherently sensitive to noise due to its discontinuous control, especially under mismatched disturbances. To improve the noise robustness and disturbance rejection of SMC, this work proposes a tightly integrated framework for DC-DC converters. Rather than simple combinations of separate modules, it seamlessly unifies a filter, disturbance observer (DOB), and SMC. A novel discrete-time interconnected observer, fusing a DOB and an adaptive Kalman filter, simultaneously reconstructs disturbances and suppresses noise. By leveraging filtered states and disturbance estimates, compensation for both mismatched disturbance and input discretization dynamics is explicitly embedded into an improved sliding mode surface. Rigorous stability analysis of the closed-loop system is provided, and comparative simulations and experiments on DC-DC converter are conducted against three benchmark control methods.

Ensuring both safety and stability remains a fundamental challenge in learning-based control, where goal-oriented policies often neglect system constraints and closed-loop state convergence. To address this limitation, this paper introduces the Predictive Safety–Stability Filter (PS2F), a unified predictive filter framework that guarantees constraint satisfaction and asymptotic stability within a single architecture. The PS2F framework comprises two cascaded optimal control problems: a nominal model predictive control (MPC) layer that serves solely as a copilot, implicitly defining a Lyapunov function and generating safety- and stability-certified predicted trajectories, and a secondary filtering layer that adjusts external command to remain within a provably safe and stable region. This cascaded structure enables PS2F to inherit the theoretical guarantees of nominal MPC while accommodating goal-oriented external commands. Rigorous analysis establishes recursive feasibility and asymptotic stability of the closed-loop system without introducing additional conservatism beyond that associated with the nominal MPC. Furthermore, a time-varying parameterisation allows PS2F to transition smoothly between safety-prioritised and stability-oriented operation modes, providing a principled mechanism for balancing exploration and exploitation. The effectiveness of the proposed framework is demonstrated through comparative numerical experiments.

SP-based synthesis yields two-time-scale control that allows compliant-joint robots to achieve high-quality tracking at low implementation cost. Composite learning enables exact online identification and control of robots without the stringent condition known as persistent excitation (PE). However, to achieve exact online identification for compliant-joint robots, parameter update derived from SP-based synthesis and composite learning requires physically unavailable states. This paper presents a novel SP-based composite learning robot control (SP-CLRC) strategy for compliant-joint robots that achieves exact online identification and control without requiring access to physically unavailable states. In the proposed method, link-side and actuator-side parameters are estimated separately, enabling exact online identification using available robot states. A two-time-scale composite learning method is proposed to guarantee practical exponential stability of the closed-loop system with parameter convergence under interval excitation, a condition strictly weaker than PE. Experiments on a two-degree-of-freedom robot driven by series elastic actuators have shown that the proposed SP-CLRC significantly outperforms the baseline in online identification and tracking accuracy.

Dear Editor, This letter deals with model predictive control (MPC) of disturbed underactuated overhead crane systems under system uncertainties and disturbances. Considering the reliance on high-precision prediction model of MPC, we enhance first-principle dynamics with a residual term representing unmodeled dynamics. The residual, learned by a Gaussian process (GP) offline, improves training efficiency, maintains physical interpretability and enhances generalization. We then integrate a generalized extended state observer (ESO) for real-time disturbance estimation and compensation, ensuring safe bridge crane operation. Comparative experiments confirm that the proposed data-physics hybrid MPC substantially improves dynamic response and anti-disturbance performance.

Accurate localization is crucial for mobile robots operating in complex operational environments, yet the standard Unscented Kalman Filter (UKF) suffers from significant performance degradation when exposed to non-zeromean or time-varying disturbances that violate its modeling assumptions. To address this challenge, we propose an Extended-State Unscented Kalman Filter (ESUKF) that augments the bicycle-model state vector with multiplicative disturbance parameters acting on translational and steering dynamics. This formulation enables joint estimation of robot states and disturbance factors, allowing the filter to explicitly compensate for model mismatch caused by friction variations, terrain slope changes, and actuator imperfections. A simulation framework is developed to evaluate the proposed method under multiple disturbance regimes. Comparative results show that ESUKF consistently outperforms the standard UKF, yielding lower trajectory and orientation estimation errors and markedly reduced estimation bias. The disturbance estimates converge accurately once adequate system excitation is present, demonstrating the effectiveness and robustness of the augmented-state filtering approach. These findings confirm ESUKF as a disturbance-resilient localization framework suitable for mobile robot navigation in uncertain environments

Model Predictive Path Integral (MPPI) controllers are drawing increasing attention for their ability to efficiently handle complex systems by leveraging GPU acceleration while with flexible prediction models and cost functions. However, their performance generally degrades with low-quality prediction models and unknown external disturbances. Existing methods that rely solely on feedforward disturbance compensation are limited by the assumption of matched disturbances, which rarely holds in practice due to the complex lumped disturbances. To this end, we propose a novel Disturbance-Aware (DA-) MPPI framework, which seamlessly integrates an Extended high-order Sliding Mode Observer (ESMO) into MPPI. The ESMO provides accurate estimates of uncertainties and external disturbances, which are directly incorporated into the MPPI rolling dynamics to improve prediction and therefore tracking control performance. The proposed algorithm is verified against the baseline MPPI in AirSim simulation environment by stochastic simulation. Comparatively statistical experiments show that incorporating ESMO within the MPPI framework significantly enhances tracking performance, with the RMSE reduction in term of mean by 8.0%, 17.7%, 6.17%, 12.9% and in term of standard variance by 11.5%, 26.0%, 10.4%, and 9.2% in four representative scenarios. The effects of target velocity and prediction horizon on control performance are also systematically evaluated. These results validate the robustness and accuracy of the DA-MPPI controller in complex and uncertain environments.

Sensorless control for permanent magnet synchronous motor (PMSM) is drawing increasing research interest in both academia and industry. To enhance the robustness of sensorless control schemes, disturbance observers are typically designed for feedforward disturbance compensation. However, the conventional framework, which separately estimates speed/position and disturbances, may lead to system instability, particularly when disturbances significantly affect speed/position estimation performance. To address this issue, this article proposes a novel estimation-based control framework that simultaneously estimates system states and disturbances. Within this framework, two simultaneous observers are introduced. First, an augmented extended Kalman filter that extends disturbances as additional states and second, a composite extended Kalman filter integrating disturbance observer and extended Kalman filter. The feasibility and effectiveness of the proposed sensorless control schemes are validated through extensive comparative hardware experiments. Compared with the conventional separate-estimation framework, the proposed approach significantly improves speed and position-estimation accuracy under load disturbances in sensorless PMSM drives.

Grapevine Leafroll Disease (GLD) poses a significant economic burden on the wine industry in major wine-producing regions. Conventional methods of phenotyping GLD are inefficient and delay vineyard management decisions. The emergence of affordable Unmanned Aerial Vehicles (UAV) provides unprecedented opportunities for GLD high-throughput phenotyping. However, detecting GLD-infected grapevines at the canopy level using UAV images is still a challenge due to the subtle differences between GLD canopy features and background. In this paper, we propose a GLD Detector (GLDD) for mapping GLD epidemics from UAV images. A new attention mechanism module, namely, Channel Attention with Transformers (CAT) is proposed to alleviate the difficulty of extracting high-resolution features from the elongated canopy. We redesigned YOLOv7-tiny for GLDD and conducted a series of ablation experiments to evaluate its performance. Experimental results show that GLDD outperforms YOLOv7-tiny by 3.1% and YOLOv6-tiny by 6.5%, reaching an accuracy of 88.2%. In comparison to several one-stage object detectors such as YOLOv5, YOLOX, PP-YOLOE, and YOLO-FaceV2, GLDD obtains the best detection results. Additionally, compared to convolutional-based detectors such as Faster-RCNN and transformer-based detector SwinT, GLDD performs better in speed and accuracy. Furthermore, GLD-infected grapevine distribution is also mapped by using GLDD detection results at the field scale.

Accurate modeling of dynamic systems is essential for robotics, enhancing system perception and control performance. This work tackles causal modeling challenges for mobile robots under complex uncertainties, including internal model inaccuracies and external environmental disturbances. Unlike first-principle or purely data-driven methods, we propose Pet-NODE, an advanced Neural Ordinary Differential Equation (NODE) framework that integrates physical priors with temporal features for high-fidelity system modeling. To further embed domain knowledge, we introduce a novel loss function with self-prediction objectives, ensuring adherence to physical principles. Extensive experiment evaluations, including ablation studies and comparisons against Nominal model, K-NODE and PI-TCN methods, demonstrate Pet-NODE’s robustness, interpretability, and superior localization accuracy on a self-collected wheeled robot dataset.

Belt deviation in circular pipe conveyor systems could lead to material spillage, environmental contamination, reduced efficiency, and accelerated belt wear. Real-time belt deviation detection is crucial for ensuring safe and efficient operation. However, existing methods, primarily developed for flat-belt systems, heavily rely on distinct belt-background semantics, making them unsuitable for circular conveyors where indistinct target features, motion blur, and occlusions pose significant challenges. To address these issues, we reformulate the deviation detection task as an object detection problem, proposing BeltLineNet, a lightweight, real-time deviation detection network specifically designed for circular pipe conveyors. Our method integrates explicit feature learning with a coarse-to-fine global feature fusion mechanism, enhancing belt line representation under complex conditions. Additionally, a shape-prior loss strategy is introduced to improve supervision during training, ensuring more accurate detection of elongated targets. The model is also optimized for real-time deployment through layer-adaptive pruning, achieving a balance between accuracy and computational efficiency. Extensive comparative experiments involving 8 state-of-the-art (SOTA) feature extraction networks and 6 feature enhancement strategies on a self-collected dataset under industrial conditions demonstrate that BeltLineNet surpasses SOTA feature extraction and fusion networks, improving average precision (AP) by 8.23% before pruning and 6.33% after pruning.