Tactile Perception and Control of a Soft Shear-Sensitive Optical Tactile Sensor

Student thesis: Doctoral ThesisDoctor of Philosophy (PhD)


Humans perform complex manipulation and contact-based interactions in their day to day endeavours relying heavily on their sense of touch to perform these ordinary yet complex interactions with their environment. Consequently, robots should also be endowed with this sensing modality if we expect them to function well in a complex unstructured environment and be useful in various settings such as industrial and assisted living.

The goal of this work is to implement a tactile perception and control framework to enable robots to interact with their environment. We start by providing a way to visualise high-dimensional tactile data which facilitates the understanding of the behaviour of a touch sensor in contact experiments. We use this as a foundation for a regression method to output continuous-valued tactile feedback, which can be easily integrated into a controller. We start by considering a discrete-contact scenario, and then we extend our work for a continuous-contact scenario which makes perception more challenging due to its path dependence. We explain the physical meaning of the derived low-dimensional tactile data representation and show how it relates to shear-invariance for the continuous-contact case. We successfully demonstrate our perception system by integrating it with a simple controller that drives the robot to perform contour following of various objects. Finally, we consider dynamic tasks where an object is moved externally. We achieve an object-following behaviour by implementing a controller that uses shear tactile feedback and we successfully achieve a complex behaviour where the robot traces the contour of a moving object by using a switching controller.

The method proposed in our work is interpretable and shows generalisation capabilities and we expect it can be used in more complex scenarios than those currently considered, thus increasing the robots' capabilities to interact with the environment.
Date of Award24 Jun 2021
Original languageEnglish
Awarding Institution
  • The University of Bristol
SupervisorDavid A W Barton (Supervisor), Nathan F Lepora (Supervisor) & Arthur G Richards (Supervisor)

Cite this