Learning-based modelling and design of Electrical impedance tomography (EIT) enabled soft tactile skins

Project ID: 2228cd1292 (You will need this ID for your application)

Research Theme: Artificial Intelligence and Robotics

Lead Supervisor: Thomas George Thuruthel

Project Summary:

The human hand has around 17000 tactile units that provide rich contact information vital for everyday object manipulation. Current robotic systems still rely only on visual data for the same task. There is a growing consensus that touch information aided with soft interactions is vital for dexterous manipulation. The primary objective of this project is to conceive and create soft tactile skins that can match the human skin in terms of both sensory capabilities and physical properties. Electrical impedance tomography is a technique for inferring a tomographic image of a body by measuring electrical conductivity across the part of an object using surface electrode measurements. By combining EIT technology with a soft conductive material, a sensorized robotic skin can be developed. There are several advantages of these electronic skins over other designs. First, all electronic components and electrodes can be located away from the sensing area, ensuring that the sensing surface remains entirely soft. Second, the number of measurements scale to the power of the number of electrodes. This project is an extension of a recent Royal Society starting grant that successfully validated this concept. The DTP proposal is to expand this work to design and model these complex nonlinear sensory skins. There are several research questions to be investigated in the beginning of the project.

The choice of conductive material to be used and its properties for optimal sensing.
The shape and morphology of the conductive skin and the placement of electrodes
Interpretation of the sensor data and estimate its physical cause with high accuracy using learning-based approaches.