Postgraduate research students

Tactile sensors play a crucial role in enabling machines, robots and prosthetics to perceive and interact with the environment through touch. They are used to detect physical properties such as pressure, force, texture, etc, by mimicking the human sense of touch. Therefore, the surface of such sensor must be made of pressure sensitive materials, such as piezoresistive ones.

Electrical Impedance Tomography (EIT) is a new technology being implemented in tactile sensors to visually represent the changes in conductivity measured by electrodes placed only at the boundaries of the sensor. The main novelty of this technology is that it leaves the surface of the sensor free from wires (which reduces the chances of the sensor breaking).  Deep learning algorithms, such as Convolutional Neural Networks (CNNs) are used to enhance the reconstruction process of those images that map and represent the magnitude of the touched points.

However, researchers have only proven the success of implementing EIT in tactile technology by training such algorithms with simulation data, which is not real enough. In order to use it in real world applications, we are going to build our own EIT tactile sensor using soft and stretchable piezoresistive materials. By acquiring data from such sensor, we will train different CNN models. Moreover, we also intend to reduce the time display that researchers have previously highlighted, as well as making the model more robust to noise.  

On the other hand, traditional EIT sensors have been studied on flat surfaces, regardless the simulation or real world. Therefore, we will also evaluate the performance of such over irregular surfaces (such as a cylinder), as I will be using soft and stretchable materials.

Implementing this EIT technology into tactile sensors would have a huge impact on understanding robotic manipulation, improving prosthetic feedback, and easing up medical diagnosis, offering a closer mimicking of human touch sensation.