Online Morphological Adaptation for Tactile Sensing Augmentation.

Area of Science:

• Robotics
• Materials Science
• Biomimicry

Background:

• Sensor morphology significantly impacts tactile sensing performance, offering enhanced sensitivity and computational capabilities.
• Designing optimal sensor structures for diverse tactile tasks is challenging due to varying morphology requirements.
• The ability to dynamically alter sensor morphology is crucial for adapting to different sensing needs.

Purpose of the Study:

• To introduce a jamming filter system for dynamically altering tactile sensor morphology.
• To investigate the relationship between sensor structure variations and information gain in tactile sensing.
• To quantify the impact of adaptive morphology on object classification accuracy.

Main Methods:

• Development of a jamming filter adaptable to various tactile sensors.
• Experimental analysis of information gain from sensors with varied morphologies.
• Evaluation of object classification accuracy using standard classifiers with and without the jamming filter.

Main Results:

• The jamming filter enables online shaping and molding of tactile sensor structures.
• Changes in sensor morphology demonstrably alter the information acquired during tactile sensing tasks.
• Appropriate sensor morphology significantly influences discrimination capabilities.
• Utilizing the jamming filter increased object classification accuracy by up to 28%.

Conclusions:

• Dynamically adjustable sensor morphology is a viable approach to enhance tactile sensing.
• The jamming filter provides a flexible solution for optimizing sensor structure for specific tasks.
• Adaptive morphology is key to improving the performance of tactile sensors in object recognition and classification.