Biomimetic Intelligence and Robotics最新文献

Cervical spondylotic myelopathy (CSM) is the main cause of adult spinal cord dysfunction, mostly appearing in middle-aged and elderly patients. Currently, the diagnosis of this condition depends mainly on the available imaging tools such as X-ray, computed tomography and magnetic resonance imaging (MRI), of which MRI is the gold standard for clinical diagnosis. However, MRI data cannot clearly demonstrate the dynamic characteristics of CSM, and the overall process is far from cost-efficient. Therefore, this study proposes a new method using multiple gait parameters and shallow classifiers to dynamically detect the occurrence of CSM. In the present study, 45 patients with CSM and 45 age-matched asymptomatic healthy controls (HCs) were recruited, and a three-dimensional (3D) motion capture system was utilized to capture the locomotion data. Furthermore, 63 spatiotemporal, kinematic, and nonlinear parameters were extracted, including lower limb joint angles in the sagittal, coronal, and transverse planes. Then, the Shapley Additive exPlanations (SHAP) value was utilized for feature selection and reduction of the dimensionality of features, and five traditional shallow classifiers, including support vector machine (SVM), logistic regression (LR), k-nearest neighbor (KNN), decision tree (DT), and random forest (RF), were used to classify gait patterns between CSM patients and HCs. On the basis of the 10-fold cross-validation method, the highest average accuracy was achieved by SVM (95.56%). Our results demonstrated that the proposed method could effectively detect CSM and thus serve as an automated auxiliary tool for the clinical diagnosis of CSM.

Variable Impedance control allows robots and humans to safely and efficiently interact with unknown external environments. This tutorial introduces online impedance adaptation control (OIAC) for variable compliant joint motions in a range of control tasks: rapid ($<1s$) movement control (i.e., whipping to hit), arm and finger impedance quantification, multifunctional exoskeleton control, and robot-inspired human arm control hypothesis. The OIAC has been introduced as a feedback control, which can be integrated into a feedforward control, e.g., learned by data-driven methods. This integration facilitates the understanding of human and robot arm control, closing a research loop between biomechanics and robotics. It shows not only a research way from biomechanics to robotics, but also another reserved one. This tutorial aims at presenting research examples and Python codes for advancing the understanding of variable impedance adaptation in human and robot motor control. It contributes to the state-of-the-art by providing an online impedance adaptation controller for wearable robots (i.e., exoskeletons) which can be used in robotic and biomechanical applications.

In recent years, the use of inertial measurement unit (IMU)-based motion capture (Mocap) systems in rehabilitation has grown significantly. This paper aimed to provide an overview of current IMU-based Mocap system designs in the field of rehabilitation, explore the specific applications and implementation of these systems, and discuss potential future developments considering sensor limitations. For this review, a systematic literature search was conducted using Scopus, IEEE Xplore, PubMed, and Web of Science from 2013 to 2022. A total of 65 studies were included and analyzed based on their rehabilitation application, target population, and system deployment and measurement. The proportion of rehabilitation assessment, training, and both were 82%, 12%, and 6% respectively. The results showed that primary focus of the studies was stroke that was one of the most commonly studied pathological disease. Additionally, general rehabilitation without targeting a specific pathology was also examined widely, with a particular emphasis on gait analysis. The most common sensor configuration for gait analysis was two IMUs measuring spatiotemporal parameters of the lower limb. However, the lack of training applications and upper limb studies could be attributed to the limited battery life and sensor drift. To address this issue, the use of low-power chips and low-consumption transmission pathways was a potential way to extend usage time for long-term training. Furthermore, we suggest the development of a highly integrated multi-modal system with sensor fusion.

Terrain classification information is of great significance for legged robots to traverse various terrains. Therefore, this communication presents an online terrain classification framework for legged robots, utilizing the acoustic signals produced during locomotion. The Mel-Frequency Cepstral Coefficient (MFCC) feature vectors are extracted from the acoustic data recorded by an on-board microphone. Then the Gaussian mixture models (GMMs) are used to classify the MFCC features into different terrain type categories. The proposed framework was validated on a quadruped robot. Overall, our investigations achieved a classification time-resolution of 1 s when the robot trotted over three kinds of terrains, thus recording a comprehensive success rate of 92.7%.

Wearable sensors for activity monitoring currently are being designed and developed, driven by an increasing demand in health care for noninvasive patient monitoring and rehabilitation training. This article reviews state-of-the-art wearable sensors for activity monitoring and motion control. Different technologies, including electromechanical, bioelectrical, and biomechanical sensors, are reviewed, along with their broad applications. Moreover, an overview of existing commercial wearable products and the computation methods for motion analysis are provided. Future research issues are identified and discussed.

Scientific advances in drone design have enabled a wide range of services, underpinned by different drones that have various aerodynamic performance. However, research to date is mostly limited focusing on conventional drones. Unconventional drones such as biomimetic drones attracted much attention due to their advantages, including precise point accessibility, altitude manoeuvrability, and no topography restriction for landing. To model the flight dynamics for biomimetic drones, the modelling accuracy is the key indicator to be determined; thus, it requires further analysis. After reviewing previous research, this paper develops a more accurate model by using appropriate methods for error mitigation. To reflect the flapping pattern of biomimetic drones, this model adopts an advanced numerical method (i.e., a quasi-steady model) to calculate their aerodynamics. The aerodynamics is also affected by the wind (acting on the drone), determined via wind-generated lift and drag terms. Therefore, this paper develops a combined aerodynamic and wind model applicable to biomimetic drones including flapping-wing drones with the following contributions. Comparative analysis discovers that the difference of drones is unsteady flows; thus, a rigorous physical model is built for flow modelling, and its novelty is a quasi-steady method to realistically quantify drone aerodynamics and wind influence. This model is demonstrated by a valid case study of the most stringent application in relation to the motion of a novel flapping-wing drone. The motion simulation of such drone is performed, and then a three-dimensional engineering prototype is built for flight test validation. This case study is implemented and the modelling performance in terms of accuracy is quantified, validating that the new model increases modelling accuracy based on research to date.

Surface electromyography (sEMG) control interface is a common method for human-centered robotics. Researchers have frequently improved the recognition accuracy of sEMG through multichannel or high-precision signal acquisition devices. However, this increases the cost and complexity of the control system. Therefore, this study developed a control interface based on the sEMG enveloped signal for a collaborative wearable robot to improve the accuracy of sEMG recognition based on the time-domain (TD) features. Specifically, an acquisition device is developed to obtain the sEMG envelope signal, and 11 types of TD features are extracted from the sEMG envelope signal acquired from the upper limb. Furthermore, a dimension reduction method based on the correlation coefficient is proposed, transforming the 11-dimensional feature into a five-dimensional envelope feature set without decreasing the accuracy. Moreover, a recognition algorithm based on a neural network has also been proposed for gesture classification. Finally, the recognition accuracy of the proposed method, principal component analysis (PCA) feature set, and Hudgins TD feature set is compared, with their accuracy at 84.39%, 72.44%, and 70.89%, respectively. Therefore, the results indicate that the envelope feature set performs better than the common gesture recognition method based on signal channel sEMG envelope signal.

The physical compliance of interaction is an important requirement for safe and efficient collaboration between robots and humans, and the realization of human–robot compliance requires robot joints with variable stiffness similar to those of human joints. In this study, based on the tissue structure and driving principle of the human arm muscle ligament, a robot joint with variable stiffness is designed, consisting of an elastic belt and serial elastic actuator in parallel. The variable stiffness of the joint is realized by adjusting the tension length of the elastic belt. Surface electromyography (sEMG) signals of the human arm are used as the characterization quantity of joint stiffness to establish the pseudo-stiffness model of the elbow joint. The stiffness of the robot joints is adjusted in real-time to match the human arm stiffness based on the changes in sEMG signals of the human arm during operation. Real-time compliant interaction of human–robot collaboration is realized based on an end stiffness matching strategy. Additionally, to verify the effectiveness of the human joint stiffness matching-based compliance control strategy, a human–robot cooperative lifting experiment was designed. The bionic variable stiffness joint shows good stiffness adjustment, and the human–robot joint stiffness matching strategy based on human sEMG signals can improve the effectiveness and comfort of human–robot collaboration.