The fast-growing techniques of measuring and fusing multi-modal biomedical signals enable advanced motor intent decoding schemes of lower-limb exoskeletons, meeting the increasing demand for rehabilitative or assistive applications of take-home healthcare. Challenges of exoskeletons’ motor intent decoding schemes remain in making a continuous prediction to compensate for the hysteretic response caused by mechanical transmission. In this paper, we solve this problem by proposing an ahead-of-time continuous prediction of lower-limb kinematics, with the prediction of knee angles during level walking as a case study. Firstly, an end-to-end kinematics prediction network(KinPreNet), ¹ consisting of a feature extractor and an angle predictor, is proposed and experimentally compared with features and methods traditionally used in ahead-of-time prediction of gait phases. Secondly, inspired by the electromechanical delay(EMD), we further explore our algorithm’s capability of compensating response delay of mechanical transmission by validating the performance of the different sections of prediction time. And we experimentally reveal the time boundary of compensating the hysteretic response. Thirdly, a comparison of employing EMG signals or not is performed to reveal the EMG and kinematic signals’ collaborated contributions to the continuous prediction. During the experiments, EMG signals of nine muscles and knee angles calculated from inertial measurement unit (IMU) signals are recorded from ten healthy subjects. Our algorithm can predict knee angles with the averaged RMSE of 3.98 deg which is better than the 15.95-deg averaged RMSE of utilizing the traditional methods of ahead-of-time prediction. The best prediction time is in the interval of 27ms and 108ms. To the best of our knowledge, this is the first study of continuously predicting lower-limb kinematics in an ahead-of-time manner based on the electromechanical delay (EMD).

The knee has gradually become an important research target for the lower extremity exoskeleton. However, the issue that whether the flexion-assisted profile based on the contractile element (CE) is effective throughout the gait is still a research gap. In this study, we first analyze the effective flexion-assisted method through the passive element’s (PE) energy storage and release mechanism. Specifically, ensuring assisting within an entire joint power period and the human’s active movement is a prerequisite for the CE-based flexion-assisted method. Second, we design the enhanced adaptive oscillator (EAO) to ensure the human’s active movement and the integrity of the assistance profile. Third, a fundamental frequency estimation based on discrete Fourier transform (DFT) is proposed to shorten the convergence time of EAO significantly. The finite state machine (FSM) is designed to improve the stability and practicality of EAO. Finally, we demonstrate the effectiveness of the prerequisite condition for the CE-based flexion-assisted method by using electromyography (EMG) and metabolic indicators in experiments. In particular, for the knee joint, CE-based flexion assistance should be within an entire joint power period rather than just in the negative power phase. Ensuring the human’s active movement will also significantly reduce the activation of antagonistic muscles. This study will aid in designing assistive methods from the perspective of natural human actuation and apply the EAO to the human-exoskeleton system.

Multiple sensors system has been adopted in most existing gait phase research. In this paper, we present a novel method to detect and discriminate gait phase based on single Inertial Measurement Unit (IMU). At first, we built a data acquisition system consisting of single IMU measuring the kinematic data of shank and foot switches labeling the data to collect the gait samples. And then we developed a gait phase recognition algorithm based on long short-term memory (LSTM) and trained it with phase-labeled data. The experimental results show that the algorithm has 91.4% accuracy on the testing set. Compared with existing approaches, the proposed method ensures the accuracy of gait phase detection and avoids the complex sensor system.

Inertial measurement units (IMUs) increasingly function as a basic component of a wearable sensor network (WSN). The IMU-based joint angle estimation (JAE) is a typical usage of IMUs for WSN, with extensive applications. However, the issue that IMUs move with respect to their original placement during JAE is still a research gap and limits the robustness of deploying the technique in real-world application scenarios. In this study, we propose to detect and correct the IMU movement online in a computationally lightweight manner. In particular, we first experimentally investigate the influence of IMU movements. Second, we design the metrics for detecting IMU movements by mathematically formulating how the IMU movement affects the IMU’s measurements. Third, we determine the optimal thresholds of metrics by synthetic IMU data from a significantly amended simulation model. Finally, a correction method is proposed to correct the effects of IMU movements. We demonstrate our method on both synthetic data and real-user data. Our proposed method achieves a detection rate of 94.8%, a misdetection rate of 9.8%, and the calculation time of 2.15 ms for detecting IMU movements and can restore the accuracy of JAE. All of the results significantly outperform the intuitive method. The results demonstrate our method is a promising solution to detect and correct IMU movements during JAE. This study will aid the deployment of IMU-based JAE in real-world applications.

Inertial measurement unit (IMU)-based lower-limb joint angle estimation (JAE) that enables wearable and real-time monitoring of kinematic and biomechanical states has been studied for decades. However, 3-degree-of-freedom (DoF) JAE of lower-limb joints still requires cumbersome calibration procedures. In this article, we propose a self-aligned method by formulating a universal 3-DoF kinematic constraint and solving it by dividing it into three subproblems. Particularly, we divide the issue of optimizing the 3-DoF kinematic constraint by: 1) estimating the main axis following a widely adopted hinge-joint assumption; 2) unifying the reference frames according to a comprehensive metric; and 3) estimating the rest two joint axes. By estimating the coordinates of the three rotational axes of the 3-DoF joint, we further estimate the 3-DoF joint angles by decoupling the joint’s rotations into rotations around the axes. We evaluate our method by accuracy comparison with other methods, test–retest reliability, and ablation experiments. The performance demonstrates that the RMSEs of our method are below 2.9° for hip and knee, below 2.6° for ankle, which are comparable with other methods and able to estimate 3-DoF angles in a self-aligned manner. The agreement between test and retest is good, and the ablation experiments demonstrate the effectiveness of sequentially solving the three subproblems. The proposed method is feasible for self-aligned JAE of 3-DoF lower-limb joints. This study may aid the wider application of IMU-based JAE in field conditions and provide a powerful component for wearable human intent decoding.

The high complexity and wide applications of Cyber-Physical Systems (CPSs) pose a large requirement on both accuracy and interpretability of the time-series anomaly detection algorithms. While a large number of deep learning algorithms have achieved excellent accuracy, the interpretability is often limited, especially when considering retaining correlations in multivariate time-series. In this paper, we propose a novel multivariate time-series anomaly detection method based on adaptive masking mechanism to improve both accuracy and interpretability, which contains a specially designed series saliency module. For more intuitive and interpretable results, a learnable adaptive mask is introduced in the series saliency module, which can disclose the influence on anomalies in both feature and temporal dimensions. The original time-series and their versions with adaptive perturbations added are then mixed via the mask forming an adaptive data augmentation method to improve the accuracy of anomaly detection. Furthermore, the anomaly detection module is model-agnostic, whether based on forecasting or reconstruction. The optimization of the training objectives will lead to more accurate and interpretable detection results. With four real-world datasets, we demonstrate that the adaptive mask can provide more accurate anomaly detection results with meaningful interpretations in the form of a mask matrix.