Musculoskeletal models have been widely used for detailed biomechanical analysis to characterise various functional impairments given their ability to estimate movement variables (i.e., muscle forces and joint moments) which cannot be readily measured in vivo. Physics-based computational neuromusculoskeletal models can interpret the dynamic interaction between neural drive to muscles, muscle dynamics, body and joint kinematics and kinetics. Still, such set of solutions suffers from slowness, especially for the complex models, hindering the utility in real-time applications. In recent years, datadriven methods have emerged as a promising alternative due to the benefits in speedy and simple implementation, but they cannot reflect the underlying neuromechanical processes. This paper proposes a physics-informed deep learning framework for musculoskeletal modelling, where physics-based domain knowledge is brought into the data-driven model as soft constraints to penalise/regularise the data-driven model. We use the synchronous muscle forces and joint kinematics prediction from surface electromyogram (sEMG) as the exemplar to illustrate the proposed framework. Convolutional neural network (CNN) is employed as the deep neural network to implement the proposed framework. Simultaneously, the physics law between muscle forces and joint kinematics is used the soft constraint. Experimental validations on two groups of data, including one benchmark dataset and one self-collected dataset from six healthy subjects, are performed. The experimental results demonstrate the effectiveness and robustness of the proposed framework.
Myoelectric control has gained much attention which translates the human intentions into control commands for exoskeletons. The electromyogram (EMG)-driven musculoskeletal (MSK) model shows prominent performance given its ability to interpret the underlying neuromechanical processes among the musculoskeletal system. This model-based scheme contains inherent physiological parameters, e.g., isometric muscle force, tendon slack length, or optimal muscle fibre length, which need to be tailored for each individual via minimising the differences between the experimental measurement and model estimation. However, the creation of the personalised EMGdriven MSK model through the evolutionary algorithms is timeconsuming, hurdling the use of the EMG-driven MSK model in practical scenarios. This paper proposes a computational efficient optimisation method to estimate the subject-specific physiological parameters for a wrist MSK model based on the direct collocation method. By constraining control variables to the experimentally measured EMG signals and introducing the physiological parameters into control variables, fast optimisation is achieved by identifying the discretised parameters at each grid simultaneously. Experimental evaluations on 12 healthy subjects are performed. Results demonstrate the proposed method outperforms the baseline optimisation algorithms used in the literature, including genetic algorithm, simulated annealing algorithm, and particle swarm optimisation algorithm. The proposed direct collocation method shows the possibility to alleviate the costly optimisation procedure and facilitate the use of the MSK model in practical applications.
Recent advances in WiFi-based device-free localization (DFL) mainly focus on stationary scenarios, and ignore the environmental dynamics, hindering the large-scale implementation of the DFL technique. In order to enhance the localization performance in nonstationary environments, in this paper, a novel multidomain collaborative extreme learning machine (MC-ELM)-based DFL framework is proposed. Specifically, the whole environment is first divided into several sub-domains depending on the distributions of the collected data using clustering algorithm, and a corresponding number of local DFL models are then built to represent these sub-domains separately. Finally, a global DFL model is achieved through seamlessly integrating all the local DFL models in a global optimization manner. The created MC-ELM-based DFL model also can be incrementally updated with sequentially coming data without retraining to track the environmental dynamics. Extensive experiments in several indoor environments demonstrate the robustness and generalization of the proposed MC-ELM-based DFL framework.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.