Computational time scales for social interaction in multi-scale robotics

Song, Brian Tzu-Ching (2015). Computational time scales for social interaction in multi-scale robotics MPhil Thesis, School of Information Technology and Electrical Engineering, The University of Queensland. doi:10.14264/uql.2015.1027

Attached Files (Some files may be inaccessible until you login with your UQ eSpace credentials)
Name Description MIMEType Size Downloads
s4028416_mphil_submission.pdf Thesis (open access) application/pdf 3.70MB 0

Author Song, Brian Tzu-Ching
Thesis Title Computational time scales for social interaction in multi-scale robotics
School, Centre or Institute School of Information Technology and Electrical Engineering
Institution The University of Queensland
DOI 10.14264/uql.2015.1027
Publication date 2015-11-06
Thesis type MPhil Thesis
Supervisor Janet Wiles
Daniel Angus
Total pages 144
Total colour pages 52
Total black and white pages 92
Language eng
Subjects 0801 Artificial Intelligence and Image Processing
Formatted abstract
Computational processes are becoming ever more complex and computationally expensive in modern robotics and advanced automation. Robotic social interaction is an important and developing area in modern robotics that involves complex and computationally intensive processes which operate on a range of computational time scales. One of the key challenges in social robotics is for robots to react quickly, in real-time, to maintain genuine sociability with humans. A novel solution for speeding up robotic reaction times is through the use of dynamic load distribution and parallelisation of robotic behaviours within a distributed control framework. This project aims to enhance computational performance in social robotics by developing a novel computational distribution paradigm (CDP), which is a software platform that manages load distribution and parallelisation while maintaining distributed control and dependencies of functional abilities for a mobile robot. The CDP provides dynamic load-balancing and parallelisation of processes across a multi-scale hardware hierarchy that caters for various computational time scales and also enables access to scalable, high performance computing resources, the Internet and external functional capabilities. Multi-scale and hierarchically based aspects of distributed computing are used by the CDP where trade-offs are made between computational speed, physical size and financial costs. Trade-offs, found in the design of traditional computing hardware, can also be applied to robotic cognition due to the inherent similarities of embodied computational performance and traditional computing characteristics. Despite the project’s focus towards robotic social behaviours, the CDP can be applied to computational processes running on any combination of computational hardware grouped by a network, including embodied mobile devices.

The CDP was designed and implemented as a software platform with an application programming interface using Aldebaran’s Nao robot as the robot platform. Interactive human-robot social behaviours that involved sound localisation and object recognition were implemented with the CDP using the full range of a computational hardware hierarchy to demonstrate the CDP’s functionality. The social behaviour experiments showed that distribution architectures for mobile robotic platforms were beneficial in improving the performance of robot computation and enabling increased functionality in some situations. The developed CDP solved, at the most basic levels, the problem of dynamic load distribution and parallelisation for robotic behaviours within a distributed control framework and provides a new approach to implementing robot behaviours.
Keyword Robotics
Distributed systems
Cloud computing
Cloud Robotics
Parallel computing

Document type: Thesis
Collections: UQ Theses (RHD) - Official
UQ Theses (RHD) - Open Access
Version Filter Type
Citation counts: Google Scholar Search Google Scholar
Created: Mon, 02 Nov 2015, 17:22:29 EST by Brian Song on behalf of Scholarly Communication and Digitisation Service