Related Experiment Video
Updated: Oct 3, 2025
09:41
Estimation of Contact Regions Between Hands and Objects During Human Multi-Digit Grasping
Published on: April 21, 2023
1.8K
CNN-Based Hand Grasping Prediction and Control via Postural Synergy Basis Extraction.
Quan Liu1, Mengnan Li1, Chaoyue Yin1
1School of Information Engineering, Wuhan University of Technology, Wuhan 430070, China.
Sensors (Basel, Switzerland)
|February 15, 2022
Summary
This study simplifies robotic hand control for stroke rehabilitation by predicting grasping actions using a convolutional neural network (CNN). This method accurately estimates hand movements, enabling natural robotic operation for effective recovery.
Area of Science:
- Robotics
- Neurorehabilitation
- Biomedical Engineering
Background:
- Stroke survivors often face challenges with hand function, necessitating effective rehabilitation strategies.
- Controlling robotic manipulators for hand activity training requires simplifying complex human hand biomechanics.
- Existing methods struggle with the high dimensionality of human hand motion data.
Purpose of the Study:
- To develop a method for simplifying hand grasping actions for robotic control.
- To reduce data dimensions for more efficient robot control in rehabilitation.
- To improve the naturalness and effectiveness of robotic assistance for stroke patients.
Main Methods:
- Explored relationships among hand grasping actions by extracting postural synergy basis.
- Proposed a convolutional neural network (CNN)-based method for hand activity prediction using motion data.
- Utilized predicted hand grasping actions to control a simulated robotic model.
Main Results:
- Achieved up to 94% prediction accuracy for selected hand motions.
- Demonstrated natural operation of the robotic model based on patient's movement intention.
- Successfully completed grasping tasks, facilitating active rehabilitation.
Conclusions:
- The proposed CNN-based method effectively simplifies hand grasping actions and reduces data dimensions.
- Accurate prediction of hand activity enables natural and intuitive control of robotic manipulators.
- This approach shows significant potential for enhancing stroke patient rehabilitation through robotic assistance.
