Experience
Research Associate, Rio Tinto Centre for Mine Automation
January 2017 to February 2019
- Developed probabilistic models of geological boundaries using Gaussian Processes
- Extended and verified these models though multiple real data sources
- Investigated methods of improving model accuracy and computation efficiency
- Built scalable data modelling processes using Slurm that ran on a computer cluster
- Generated and executed experimental plans to meet research objectives
- Converted high level research goals into actionable workflow plans
- Remained agile to keep pace with changing stakeholder requirements
- Worked independently whilst being part of a team
- Focused on specific research tasks whilst understanding 'the bigger picture'
- Demonstrated completed research milestones and their relation to 'the bigger picture'
- Transferred knowledge and research milestones to the tech team for implementation
- Generated reports and presentations for internal stakeholders
- Skills gained/strengthened
- Linux/bash scripting, collaboration, shared code-base programming, communication, managing stakeholder expectations
PhD (Human Robot Interaction), University of Sydney
2012 to 2016
- Formulated and executed a 4 year long research project
- Investigated the comfort of pairs of people as they were approached by a robot
- Undertaken whilst part of the Social Robotics Group at the Australian Centre for Field Robotics
- Developed a comprehensive and statistically sound data processing pipeline
- Data acquisition: Design and execution of an experiment
- Data analysis: Mann-Whitney U, Kruskal-Wallis ANOVA, Rayleigh Uniformity, and Watson’s U2 statistical tests
- Communicated results through written and verbal mediums
- Written: Submitted work to international journals and conferences, compiled complete project into a thesis document
- Verbal: Presentation of research papers at conferences, internal presentations to ACFR
- Skills gained/strengthened
- Analytical thinking, pragmatic methodology, communication, working as part of a team, and the ability to self-motivate and work independently
Masters by Research (Human Robot Interaction), University of Sydney
2010 to 2012
- Formulated and executed a 2.5 year long research project
- Investigated the potential for gait-based person recognition
- Undertaken whilst part of the Social Robotics Group at the Australian Centre for Field Robotics.
- Implemented and compared different supervised and unsupervised machine learning models
- Data acquisition: RGB-D data from a Kinect, 3D joint data from a motion-capture system
- Supervised models: SVM, Logitboost
- Unsupervised models: K-means, Affinity Propagation, Spectral Propagation
- Communicated results through written and verbal mediums
- Written: Submitted work to international conferences, compiled complete project into a thesis document
- Verbal: Presentation of research papers at conferences, internal presentations to ACFR
- Skills gained/strengthened
- Analytical thinking, pragmatic methodology, communication, working as part of a team, and the ability to self-motivate and work independently
University Tutor (Teachers Assistant), University of Sydney
2010 to 2016
Recipient of the Faculty Tutoring Award
- Assisted with the running of tutorials and laboratory sessions
- Involved preparatory work for running the classes and marking assignments
- Generated and presented class lectures
- Provided unofficial supervisory assistance to students completing undergraduate theses
- Theses were related to mechatronic engineering
- Mentored the students on a wide variety of topics not limited to: experiment design, robot design, data analysis, research skills, communication of results
Language Editor, Paladyn Journal of Behavioural Robotics
2012 to 2015
- Volunteer position to occasionally review submitted articles and assist with language editing
Education
PhD - Australian Center of Field Robotics, The University of Sydney
How Should a Robot Approach a Pair of People?
2012 to 2016
Recipient of the Australian Postgraduate Award, Research Scholarship from the Australian Government
This thesis explored how comfortable pairs of people are when approached by a mobile robot from different directions. The primary contributions of this work are quantitative analyses of of experiment results showing how group comfort differs from that of a lone individual, and how intra-group orientation changes individual comfort responses. A secondary contribution is the introduction of directional statistics to the field of HRI and demonstrating its use as an informative data analysis method.
The core of this thesis focused on the design of an experiment that would produce participant data in a form that could then be passed through non-parametric linear and directional statistic pipelines. This process involved the programming and verification of statistical tests before using the process to analyse participant data acquired from performed experiments. The results of the process then had to be communicated to relevant audiences through both presentations and technical documentation.
Masters by Research - Australian Center of Field Robotics, The University of Sydney
Human Recognition Through Gait Analysis Using an Inexpensive Visual Sensor
2010 to 2012
This thesis investigated the capability of the Kinect to recognise people based on their gait in an open-plan office environment. The experiment was then emulated with a motion capture system to help identify shortfalls with the Kinect. Machine learning techniques are employed to show that while the Kinect is outperformed by state-of-the-art sensors, there is potential for it to be used as a cheap sensor for gait-based human identification.
Bachelor of Mechatronic Engineering (Honours) - The University of Sydney
2004 to 2009
Bachelor of Science (Double Major in Advanced Mathematics) - The University of Sydney
2004 to 2009
Related Skills
Areas of Expertise
- Docker, Slurm, bash scripting
- Tensorflow, Keras, GPU Computation
- ROS
Big Data
- GCP Big Data and Machine Learning Fundamentals
- Leveraging Unstructured Data with Cloud Dataproc on GCP
- Serverless Data Analysis with Google BigQuery and Cloud Dataflow
- Serverless Machine Learning with Tensorflow on GCP
- Building Resilient Streaming Systems on GCP
Programming Languages
Operating Systems
Publications
How Should a Robot Approach Two People?
International Journal of Human-Robot Interaction 2017 (in publication).
[Pdf] [Publication Link]
Understanding Group Comfort Through Directional Statistics
International Conference on Social Robotics 2015.
[Pdf] [Publication Link]
Group vs. Individual Comfort When a Robot Approaches
International Conference on Social Robotics 2015.
[Pdf] [Publication Link]
Group Comfortability when a Robot Approaches
International Conference on Social Robotics 2014.
[Pdf] [Publication Link]
Unsupervised Clustering of People from ‘Skeleton’ Data
International Conference on HRI 2012.
[Pdf] [Publication Link]
A Comparison of Unsupervised Learning Algorithms for Gesture Clustering
International Conference on HRI 2011.
[Pdf] [Publication Link]