Electro-hydraulic actuators (EHAs) are key linear drive components in various industrial applications. This paper addresses the challenge of achieving precise displacement tracking control for EHAs with only noisy displacement measurements. We propose a novel control approach, which consists of a Kalman extended state observer (KESO) with an adaptive sliding mode controller (ASMC). First, compared to the conventional high-gain design for the extended state observer (ESO), the Kalman filtering technique is utilized to tune the ESO feedback gain, effectively mitigating observation failures caused by measurement noise. Second, to ensure stability and minimize tracking errors in sliding mode control, a switching-gain adaptation is designed based on the desired switching gain via the derivative of the sliding variable. To validate the effectiveness of the proposed approach, simulation experiments are conducted in the Amesim simulation software. The results conclusively demonstrate that the proposed KESO-ASMC achieves significantly improved trajectory tracking performance, even in the presence of measurement noise and unknown disturbances.

This study presents a discrete‐time sliding mode control (DSMC) with an input filter for a low damping electro‐hydraulic actuator (EHA) suffering from large control volume variation. Non‐switching type reaching law and one‐step delayed estimation of disturbance are combined in the DSMC. Different from existing DSMC strategies which view parametric uncertainty as a lumped bounded disturbance, the influence of parametric uncertainty on the system stability is analysed, which results to a two terms iterative expression of state. The stability criterion for this iteration is put forward. To stabilise the EHA that has a large control volume variation, a first‐order input filter for the control input is introduced. The disturbance resulting from the input filter is expressed and the stability criterion with the input filter is derived. Both simulation and experimental results show the effectiveness of the proposed approach on the position tracking of the EHA with large volume variation.

This paper proposes a novel prediction model termed the social and spatial attentive generative adversarial network (SSA-GAN). The SSA-GAN framework utilizes a generative approach, where the generator employs social attention mechanisms to accurately model social interactions among pedestrians. Unlike previous methodologies, our model utilizes comprehensive motion features as query vectors, significantly enhancing predictive performance. Additionally, spatial attention is integrated to encapsulate the interactions between pedestrians and their spatial context through semantic spatial features. Moreover, we present a novel approach for generating simulated multi-trajectory datasets using the CARLA simulator. This method circumvents the limitations inherent in existing public datasets such as UCY and ETH, particularly when evaluating multi-trajectory metrics. Our experimental findings substantiate the efficacy of the proposed SSA-GAN model in capturing the nuances of pedestrian interactions and providing accurate multimodal trajectory predictions.

Human-robot collaborative digital twin (HRC-DT) systems often suffer dynamic mismatches between virtual models and their physical counterparts due to data noise, mapping latency, system instability, and external sensor disturbances. To mitigate these issues, we propose a spatiotemporal alignment framework that fuses multi-source data resampling with adaptive feedback control. First, all sampling subsystems and the digital twin platform are time-synchronized via the Network Time Protocol (NTP). We then construct a motion-confidence-driven adaptive filter (MCDAF) to suppress abrupt distortions in human skeletal data, followed by a spatial-correlation-based predictive resampling (SCPR) module that aligns sensor streams with DT update timestamps. To reduce computational and communication load while preserving the necessary human body tracking accuracy, we introduce a dynamic sampling rate proportional-derivative (PD) controller based on human joint dynamics and human-robot distance. Finally, to promptly detect any physical disturbances that the depth camera may encounter during in operation, we developed an image-based self-check module for the depth-camera offset, along with a dynamic interference guard (DIG) mechanism, and a 6D pose detection-based rapid recalibration method. Experimental results demonstrate that the proposed approach enhances dynamic consistency compared with conventional methods, and provides a robust and efficient solution for reliable HRC-DT deployment.