Tactile Object Discrimination
The world is full of tools and objects for which the human hand seems ideally suited. Not only is the human hand a highly evolved solution to many difficult problems, but humans have used their hands to fashion a world of objects and tools suited to the strengths and weaknesses of our hands.
Here we incorporate our biomimetic sensor and strategies into modules for a mechatronic anthropomorphic robot hand to perform exploratory behaviors and discriminate objects based on sensory feedback.
Project Objective
Design and build a hand, arm, and processing system to identify objects using only tactile feedback similar to the skin on a human hand.
Physiologists have identified six exploratory strategies that humans use to identify six unique object properties by touch. We have expanded the analysis of these fundamental strategies and have developed methods to use the tactile feedback obtained from our sensors and those found in the anthropomorphic hand to mimic these exploratory movements to identify the same object features that humans are capable of detecting. Using these biomimetic principals, the system will initially undergo a training phase using these exploratory strategies to identify object properties and relate these patterns of properties to unique objects. To validate the systems capabilities to identify objects by touch, it will then be presented with unknown objects from the object set and autonomously perform these exploratory strategies and process information to classify and identify the unknown object. Overall system performance will then be quantified using the success rate of this task and the complexity of the object set.
BioTac sliding over textured surface. The vibrations produced from this interaction are sensed by the BioTac and can be used to identify the texture.