Cardiff University | Prifysgol Caerdydd ORCA
Online Research @ Cardiff 
WelshClear Cookie - decide language by browser settings

Design and development of vision-based tactile sensor for robotics applications

Rayamane, Prasad ORCID: 2023. Design and development of vision-based tactile sensor for robotics applications. PhD Thesis, Cardiff University.
Item availability restricted.

[thumbnail of Design_and_Development_of_Vision_Based_Tactile_Sensor_for_Robotics_Application_PrasadRayamane_C1750619 (1).pdf] PDF - Accepted Post-Print Version
Restricted to Repository staff only until 13 June 2025 due to copyright restrictions.

Download (36MB)
[thumbnail of Cardiff University Electronic Thesis and Dissertation Form] PDF (Cardiff University Electronic Thesis and Dissertation Form) - Supplemental Material
Restricted to Repository staff only

Download (162kB)


Tactile sensing enables humans to directly interact with the environment. In robotics, tactile sensing plays a crucial role in grasping tasks by providing additional information to complement vision, including contact forces and surface texture. Integrating tactile sensing into robots is essential to provide human-like touch capabilities. Slip detection, an essential aspect of human touch, involves perceiving relative motion between the skin and an object, and is employed by humans in various scenarios, such as determining optimal grasping strength and inferring an object’s physical properties. This thesis focuses on designing, developing and exploring novel vision-based tactile sensing technologies to enhance slip detection and manipulation capabilities in robotic systems. Firstly, a robust vision-based tactile sensor has been developed and designed, utilising computer vision techniques for precise tactile perception of objects. Secondly, a method has been proposed to enhance the mechanical characteristics of the silicone elastomer used in the sensor. This method involves applying three different coating types (latex membrane, metallic coating, and no coating) to optimise the sensor’s tactile sensitivity and performance. Furthermore, a pneumatic-based optical tactile sensor named PnuTac, has been successfully designed and developed. This sensor integrates pneumatic mechanisms and variable pressure to enhance tactile perception and manipulation capabilities. The sensor captures crucial data, including contact geometry and slip detection. When integrated with a Robotiq 2-finger gripper, the PnuTac sensor facilitates slip detection and stabilisation experiments of slipping objects, resulting in an impressive overall success rate of 87%. Additionally, a trained neural network employs real-time data from the PnuTac sensor, proficiently categorising the tool, with an overall classification accuracy of 80%. Finally, the thesis investigates the performance of the designed vision-based tactile sensor for object handover, measuring accuracy, success rate, and adaptability to various objects. The research presented in this thesis contributes to the advancement of tactile sensing technology in robotics, specifically by achieving real-time slip detection through the emulation of structures found in the human fingertip. This highlights the importance of integrating slip detection as a vital aspect in future tactile robot systems. The primary objective of this thesis is to present innovative vision-based tactile sensor designs and their applications in robotics, emphasising their robustness, spatial resolution, and variable pressure capabilities for diverse robotics applications. This thesis provides insights into the field of tactile robotics, with a specific focus on the design of vision-based tactile sensors, slip detection, tactile robot grasping, and object handover.

Item Type: Thesis (PhD)
Date Type: Completion
Status: Unpublished
Schools: Engineering
Uncontrolled Keywords: 1) Vision-Based Tactile Sensor 2) Deep Learning 3) Slip Detection 4) Object Handover
Date of First Compliant Deposit: 13 June 2024
Last Modified: 14 Jun 2024 08:08

Actions (repository staff only)

Edit Item Edit Item


Downloads per month over past year

View more statistics