Padraic Moriarty

Padraic Moriarty
Research Assistant

Qualifications:

B.Sc. Physical Optoelectronics, University of Essex 1993; H.Dip in Science in Applied Computing, IT Tralee 2013.

Research Areas of Expertise / Interest:

Software Development, Optoelectronics, Electronics

Biography:

Worked in several engineering positions in the hard drive and electronics industries. Partnered in a web solutions / graphic design company for eleven years acting as software development manager and head programmer.

Project Title:

Application of Deep Neural Networks to Computer Vision

Partner Company :

REAMDA

Project overview:

This project proposes to apply Neural Networks to autonomously land or aide in the landing of a UAV helicopter. Typically these craft land, and are launched from, the deck or a platform of a ship. In challenging weather conditions it takes a very skilled pilot to land safely. Quite often the choice is made to ditch the craft into the sea, rather than to risk a potentially destructive crash landing. This accepts a certain level or damage and inconvenience. Artificial neural networks are a machine learning technique modelled on an understanding of how the human brain works. Terms like neurons, connections and reinforcement learning are all part of the vocabulary associated with these techniques. By acquiring data from skilled pilots, the Neural Net can learn to land the UAV and eliminate human error. This should result in better success rates in landing. Furthermore, with testing and analysis of the final product, it would allow for informed decisions to be made on whether to attempt a landing or not. This research is strategically relevant to IT Tralee, the IMaR technology gateway, research Priority B (Data Analytics) and Priority N (Manufacturing Competitiveness) , is industry driven, and complements suite of projects and expertise already in place at IMaR.