Reinforcement Machine Learning in dexterous robotic end-effector systems; Fully funded PhD studentship

Currently, robots lack the ability to feel objects in the same way as humans do, which makes them unable to handle delicate objects or tell their consistencies. Although robots rely on various sensors to sense their surroundings & the objects they interact with, none are able to measure pressure/force as precisely as human skin. Wootzano Ltd has invented a novel, highly sensitive, flexible, & mass-producible e-Skin called Nanskin and is creating its own robot RoboPack with Nanskin sensors. The robot requires data generated from the Nanskin sensor to be processed via reinforcement learning algorithms. This data is processed to teach the robot how to grasp a particular object without damaging it. Furthermore, the robot has auxiliary components & the data generated from them needs to be processed as well. This project will therefore involve aspects from computer vision, electronics & flexible electronics together with machine learning to create one of its kind robotic manipulator.

The aim of this project is to develop an advanced machine & deep learning pipeline to predict whether an object will be damaged, & to consequently extract the features linked with damages. Features for the machine learning pipeline will include characteristics of the robot, as well as forces & object specifications. Unsupervised learning, including network-based & clustering methods, will be used to analyse the data collected, & to pre-train a supervised learning algorithm. Hyper-parameter optimisation & regularisation techniques will be employed where appropriate. Due to the high dimensionality of parameter spaces, multi-objective optimisation algorithms will be used to optimise robot & movement parameters, towards minimisation of the number of damaged objects. The novel algorithms & technology developed for RoboPack during this PhD project will enable never-before-possible haptic & tactile feedback for robots, prosthetics, & smart surfaces.

North East England’s universities are joining forces under a £3.9m scheme, funded by the European Regional Development Fund, to connect the region’s businesses with research to encourage growth & job creation. Teesside University is delighted to be able to offer a number of part funded industrial PhDs to eligible SMEs in the north east. This project is funded by the University & European Regional Development Funding & looks to support local firms with their research & development needs, developing new products & services in key sectors & creating high quality jobs in the local economy.

Entry Requirements
Applicants should hold or expect to obtain a relevant degree at 2.1 minimum, or an equivalent overseas degree in computer science, engineering, mathematics, physics or related subjects. Candidates with expertise in machine/deep learning are particularly welcome to apply.
International students would be subject to the standard entry criteria including English language, ATAS clearance & Tier 4 procedures. Please contact the [Email Address Removed] for further information.

About Teesside University
The School of Computing, Media and the Arts at Teesside University conducts research on a wide range of topic areas including Computer Science, Artificial Intelligence, Software Engineering, Cyber-Physical Systems, Computer Games, Animation, Media and Fine Art. In REF 2014, 69.8% of our research outputs in Computer Science and Informatics was recognised as world-leading or internationally excellent. For more information: http://www.tees.ac.uk/sections/research/computing/about.cfm

About Wootzano
Wootzano specialises in flexible electronics & has invented a novel, highly sensitive, flexible, & mass-producible e-Skin called Nanskin, find out more here: https://www.wootzano.com

How to Apply
Please apply online for this opportunity; using the PhD full time application form & ref: IIIP0003