Research

Area of Research

Rehabilitation Robotics, Haptics for Neurorehabilitation, Machine Learning & AI, Control Systems

From Sensing to Assistance: A Comprehensive Framework for Grasp Rehabilitation

The human hand is a marvel of biomechanical complexity and functional adaptability, enabling us to interact seamlessly with our environment. However, for individuals affected by neurological injuries or disorders, impaired hand function—especially the loss of grasp—can drastically reduce independence and quality of life. This research presents an end-to-end framework for grasp rehabilitation, bridging sensing, analysis, and assistance through innovative wearable technologies.

✋ Multi-Sensory Data Glove for Grasp Analysis

The first part of this work focuses on the design and implementation of a custom data glove, built using flexible resistive sensors and tailored to individual hand geometry via 3D anthropometric scanning. A comprehensive experimental study was conducted involving standardized grasp types and objects to record joint angles and fingertip forces.

Key innovations include:

• Real-time grasp classification using hybrid deep learning models (CNN-BiLSTM).

• Grasp intent prediction with transformer-based architectures.

• Force prediction models employing direct, autoregressive, and multimodal strategies.

• Grasp synergy analysis using force-angle correlations, and dimensionality reduction (PCA, t-SNE) to uncover coordination patterns among fingers.

🤖 Soft Hand Exoskeleton for Assistive Grasping

The second part introduces a tendon-driven soft hand exoskeleton designed for assistive and rehabilitative use. Developed with a modular and ergonomic structure informed by anatomical data, the exoskeleton features compliant joints and optimized tendon routing for natural motion.

Highlights include:

• Finite Element Modeling (FEM) to ensure safety and mechanical compliance.

• Data-driven NSTSM control strategy offering improved tracking accuracy and torque efficiency.

• Simulation and pilot trials validating its potential for stroke rehabilitation and prosthetic applications.

🚀 Impact and Applications

This thesis offers a unified platform that integrates multi-modal sensing, intelligent analysis, and soft robotic assistance—paving the way for advancements in:

• Neurorehabilitation

• Assistive prosthetics

• Human-robot interaction

The work contributes not just to the science of grasp recovery but also to the practical development of technologies that can enhance human capabilities and restore autonomy to individuals in need.

Work in Progress...