Subhash Pratap

This paper investigates hand grasping, a fundamental activity in daily living, by examining the forces and postures involved in the lift-and-hold phases of grasping. We introduce a novel multi-sensory data glove, integrated with resistive flex sensors and capacitive force sensors, to measure the intricate dynamics of hand movement. The study engaged five subjects to capture a comprehensive dataset that includes contact forces at the fingertips and joint angles, furnishing a detailed portrayal of grasp mechanics. Focusing on grasp synergies, our analysis delved into the quantitative relationships between the correlated forces among the fingers. By manipulating one variable at a time—either the object or the subject—our cross-sectional approach yields rich insights into the nature of grasp forces and angles. The correlation coefficients for finger pairs presented median values ranging from 0.5 to nearly 0.9, indicating varying degrees of inter-finger coordination, with the thumb-index and index-middle pairs exhibiting particularly high synergy. The findings, depicted through spider charts and correlation coefficients, reveal significant patterns of cooperative finger behavior. These insights are crucial for the advancement of hand mechanics understanding and have profound implications for the development of assistive technologies and rehabilitation devices.

Grasp synergies lead to identification of underlying patterns to develop control strategies for five-fingered prosthetic hands or exoskeletons. Data gloves play a crucial role in study of human grasping and could provide insights into grasp synergies. This paper presents the design and implementation of a data glove that has been fabricated using 3D printing technology and enhanced with instrumentation. The glove utilizes flexible sensors for the fingers and force sensors integrated into the glove at the fingertips to accurately capture grasp postures and forces. Understanding the kinematics and dynamics of human grasp including reach-to-grasp is undertaken. A comprehensive study involving 10 healthy subjects was conducted. Grasp synergy analysis is carried out to identify underlying patterns for grasping. Correlation analysis showed a strong synergy, especially between Index and Middle fingers with a 0.95 correlation coefficient. Principal Component Analysis (PCA) facilitated dimensionality reduction, revealing that three principal components capture over 97 percent of the variance in grasp postures, underscoring the complexity and synergy of hand movements.Grasp classification experiments validated the efficacy of PCA-based synergy, achieving high classification accuracies (95.84% to 92.34%) and demonstrating the method’s competitive performance in scenarios requiring reduced sensor complexity, as confirmed by confusion matrices and comparative analysis with existing methodologies. The t-SNE visualization showcased clusters of grasp postures and forces, unveiling similarities and patterns among different grasp types. These findings could serve as a comprehensive guide in design and control of five-fingered robotic hands and exoskeletons for rehabilitation applications, enabling the replication of natural hand movements.

Grasp classification is pivotal for understanding human interactions with objects, with wide-ranging applications in robotics, prosthetics, and rehabilitation. This study introduces a novel methodology utilizing a multisensory data glove to capture intricate grasp dynamics, including finger posture bending angles and fingertip forces. Our dataset comprises data collected from 10 participants engaging in grasp trials with 24 objects using the YCB object set. We evaluate classification performance under three scenarios: utilizing grasp posture alone, utilizing grasp force alone, and combining both modalities. We propose Glove-Net, a hybrid CNN-BiLSTM architecture for classifying grasp patterns within our dataset, aiming to harness the unique advantages offered by both CNNs and BiLSTM networks. This model seamlessly integrates CNNs’ spatial feature extraction capabilities with the temporal sequence learning strengths inherent in BiLSTM networks, effectively addressing the intricate dependencies present within our grasping data. Our study includes findings from an extensive ablation study aimed at optimizing model configurations and hyperparameters. We quantify and compare the classification accuracy across these scenarios: CNN achieved 88.09%, 69.38%, and 93.51% testing accuracies for posture-only, force-only, and combined data, respectively. LSTM exhibited accuracies of 86.02%, 70.52%, and 92.19% for the same scenarios. Notably, the hybrid CNN-BiLSTM proposed model demonstrated superior performance with accuracies of 90.83%, 73.12%, and 98.75% across the respective scenarios. Through rigorous numerical experimentation, our results underscore the significance of multimodal grasp classification and highlight the efficacy of the proposed hybrid Glove-Net architectures in leveraging multisensory data for precise grasp recognition. These insights advance understanding of human–machine interaction and hold promise for diverse real-world applications.

Grasping, a fundamental aspect of human motor skills, involves the intricate coordination of sensory and motor systems to manipulate objects effectively. Accurate grasp classification, crucial for advancing human-machine interaction in robotics and prosthetics, involves categorizing various hand-grasping patterns. This study explores a grasp classification method employing tactile sensors measuring fingertip grasp forces and machine learning algorithms. The study evaluates the performance of Decision Tree, Random Forest (RF), k-Nearest Neighbor (KNN), Support Vector Machine (SVM), and Neural Network classifiers in identifying and categorizing eight grasp types. The study involves the participation of 10 subjects, utilizing capacitive-based Finger TPS force sensors. The RF classifier achieves the highest accuracy at 90%, outperforming others with accuracies ranging from 71% to 87%. The results demonstrate the efficacy of data gloves and machine learning classifiers in efficient and accurate grasp identification, offering valuable insights for controlling grasping force and planning robotic motions.

The evaluation of human walking through sensor data is pivotal in diagnosing various neurological disorders. Despite the widespread use of deep learning methods to distinguish between healthy and pathological walking, the efficacy of already in-place approaches remains unclear. In this study, we have developed and implemented a Deep Neural Network (DNN) to identify healthy and pathological gaits based on a muscle sensory and knee movement dataset. The considered dataset consists of walking information from 22 subjects, with 11 classified as healthy and rest with knee pathology, sourced from the machine learning repository at the University of California, Irvine (UCI). This dataset encompasses five channels of sensor signals, with the four containing surface Electromyography (sEMG) data representing muscular initiations in lower limbs as Biceps Femoris (BF), Vastus Medialis (VM), Rectus Femoris (RF), and Semitendinosus (ST). The fifth channel comprises goniometer signals measuring the flexion/extension movements of the knee joint. To assess the classification process, two optimizers, namely Stochastic Gradient Descent (SGD) and Adaptive Moment (Adam), are considered. Subsequently, performance metrics such as precision, recall, F1-score, and confusion matrix are evaluated and discussed for both optimizers. The training accuracy in classifying healthy and pathological gaits for the two optimization algorithms is determined to be 79.83% and 90.81%, respectively. The Adam-DNN demonstrates superior performance compared to SGD-DNN in classifying healthy and pathological gaits.

Human grasps involved in activities of daily living rely on the correlation of multiple degrees of freedom in a hand. This correlation is called grasp synergies and holds promise for efficient control of robotic hands including prostheses and exoskeletons. A number of studies involving human grasping experiments have been reported. However, the approach, lift and hold phases which are intricately linked to the hand grasp for activities of daily living have hardly been part of such studies. We report a human grasping study which includes the approach, lift and hold phases. Grasp synergies in the force and posture space are identified. We have observed distinct force patterns while manipulating different objects. Additionally, our analysis of correlation coefficients between the pair of fingers highlights the cooperative behaviour and synergy between fingers while grasping. Our study contributes valuable insights into finger correlations, and synergies in terms of grasp postures and forces.