Intelligent soft robotic gripper for non-destructive grasping and attribute recognition via multi-modal waveguide...

Area of Science:

• Robotics
• Materials Science
• Sensor Technology

Background:

• Traditional mechanical grippers face limitations in handling fragile items and irregular objects.
• Electrical tactile sensors struggle with complex structures and environmental interference.
• Optical waveguide tactile sensing offers advantages but faces challenges with scale and integration.

Purpose of the Study:

• To develop a soft robotic gripper with advanced tactile sensing for intelligent grasping.
• To improve robotic manipulation capabilities in complex tasks.
• To enable accurate object attribute recognition during manipulation.

Main Methods:

• Integration of slender elastic optical waveguide sensors (EOWS) into a three-finger soft robotic gripper.
• Implementation of a closed-loop feedback control module with PID adaptive grasping.
• Application of a long short-term memory (LSTM) deep learning algorithm for object attribute recognition.

Main Results:

• EOWS demonstrated high sensitivity to bending angle, contact force, and pressure.
• The system achieved over 97% accuracy in recognizing object shape, size, and hardness.
• Experimental validation in a smart fruit-sorting system demonstrated robust and adaptive manipulation.

Conclusions:

• The developed soft robotic gripper with EOWS offers superior performance for complex manipulation tasks.
• The intelligent control and recognition system enhance robotic adaptability and precision.
• The platform shows significant potential for applications in precision agriculture, logistics, and medical robotics.