A5065533993- Neuroscience and Neural Engineering
- Advanced Memory and Neural Computing
- Tactile and Sensory Interactions
- Assistive Technology in Communication and Mobility
- EEG and Brain-Computer Interfaces
- Autism Spectrum Disorder Research
- Robotics and Sensor-Based Localization
- Video Surveillance and Tracking Methods
- Advanced Vision and Imaging
- Gaze Tracking and Assistive Technology
- CCD and CMOS Imaging Sensors
- Astronomy and Astrophysical Research
- Robotic Path Planning Algorithms
- Robot Manipulation and Learning
- Cerebral Palsy and Movement Disorders
- Stroke Rehabilitation and Recovery
- Family and Disability Support Research
- Handwritten Text Recognition Techniques
- Experimental Learning in Engineering
- Astro and Planetary Science
- Social Robot Interaction and HRI
- Digital Mental Health Interventions
- Human Mobility and Location-Based Analysis
- Smart Agriculture and AI
- IoT and GPS-based Vehicle Safety Systems
Swinburne University of Technology
2015-2024
Armstrong Atlantic State University
2020
Data61
2006-2017
Australian National University
2008-2016
National University College
2015
The University of Melbourne
2002-2014
Kaiser Permanente
2012
Ara Institute of Canterbury
2005
Napa Valley College
2003
Newcastle University
1995
Retinal visual prostheses (“bionic eyes”) have the potential to restore vision blind or profoundly vision-impaired patients. The medical bionic technology used design, manufacture and implant such is still in its relative infancy, with various technologies surgical approaches being evaluated. We hypothesised that a suprachoroidal location (between sclera choroid of eye) would provide significant safety benefits for patients, allowing them maintain preoperative residual as well gaining...
Recent work in salient object detection has considered the incorporation of depth cues from RGB-D images. In most cases, contrast is used as main feature. However, areas high background regions cause false positives for such methods, frequently contains that are highly variable depth. Here, we propose a novel saliency Local Background Enclosure (LBE) captures spread angular directions which with respect to candidate region and it part of. We show our feature improves over state-of-the-art...
We train and apply convolutional neural networks, a machine learning technique developed to learn from classify image data, Canada–France–Hawaii Telescope Legacy Survey (CFHTLS) imaging for the identification of potential strong lensing systems. An ensemble four networks was trained on images simulated galaxy–galaxy lenses. The training sets consisted total 62 406 lenses 64 673 non-lens negative examples generated with two different methodologies. applied all 171 deg2 CFHTLS wide field...
We search Dark Energy Survey (DES) Year 3 imaging data for galaxy-galaxy strong gravitational lenses using convolutional neural networks. generate 250,000 simulated at redshifts > 0.8 from which we create a set training the networks with realistic seeing, sky and shot noise. Using simulations as guide, build catalogue of 1.1 million DES sources 1.8 < g - i 5, 0.6 -r 3, r_mag 19, g_mag 20 i_mag 18.2. train two ensembles on sets consisting lenses, non-lenses, real sources. use to score images...
Translational research in vision prosthetics, gene therapy, optogenetics, stem cell and other forms of transplantation, sensory substitution is creating new therapeutic options for patients with neural blindness. The technical challenges faced by each these disciplines differ considerably, but they all face the same challenge how to assess ultra-low (ULV), who will be earliest subjects receive therapies. Historically, there were few tests ULV patients. In 1990s, field visual prosthetics...
Objective. We evaluated a novel visual representation for current and near-term prosthetic vision. Augmented depth emphasizes ground obstacles floor-wall boundaries in depth-based representation. This is achieved by artificially increasing contrast between the surface via plane extraction algorithm specifically designed to preserve low-contrast ground-surface boundaries. Approach. The effectiveness of augmented was examined human mobility trials compared against standard intensity-based...
One strategy to improve the effectiveness of prosthetic vision devices is process incoming images ensure that key information can be perceived by user. This paper presents first comprehensive results function testing for a suprachoroidal retinal device utilizing 20 stimulating electrodes. Further, we investigate whether using image filtering on light localization task implanted participants compared minimal processing. No controlled participant studies have yet investigated processing...
We present a comparison of four optical flow methods and three spatio-temporal filters for mobile robot navigation in corridor-like environments. Previous comparisons have evaluated performance only terms accuracy and/or efficiency, typically isolation. These are inadequate addressing applicability to continuous, real-time operation as part control loop. emphasise the need that consider context system, confirmed by in-system results. To this end, we give results on off-board trials two...
There is increasing interest in using computer vision and machine learning to enhance human decision making with computer-mediated assistive systems. In particular, retinal implants are a rapidly advancing technology offering individuals suffering loss due dystrophies, an opportunity restore partial vision. However, the visual representations achievable current near-term severely limited resolution contrast, placing high importance on selection of features convey via implant. Using...
We present a robust strategy for docking mobile robot in close proximity with an upright surface using optical flow field divergence and proportional feedback control. Unlike previous approaches, we achieve this without the need explicit segmentation of features image, complete gradient-based estimation (i.e., no affine models) computation. A key contribution is development algorithm to compute divergence, or time-to-contact, manner that small rotations during ego-motion. This done by...
Socially assistive robots (SARs) offer great promise for improving outcomes in paediatric rehabilitation. However, the design of software and interactive capabilities SARs must be carefully considered context their intended clinical use. While previous work has explored specific roles functionalities to support rehabilitation, few have such ongoing deployment. In this article, we present a two-phase situ process health care, emphasising stakeholder engagement on-site development. We explore...
We search Dark Energy Survey (DES) Year 3 imaging for galaxy-galaxy strong gravitational lenses using convolutional neural networks, extending previous work with new training sets and covering a wider range of redshifts colors. train two networks images simulated lenses, then use them to score postage stamp 7.9 million sources from the chosen have plausible lens colors based on simulations. examine 1175 highest-scored candidates identify 152 probable or definite lenses. Examining an...
Purpose: Socially assistive robots are emerging as a method of supporting the rehabilitation children with physical disabilities. To date there has been no in-depth analysis parent and child perspectives regarding use socially for pediatric rehabilitation. The purpose this study was to capture experiences parents who participated in session robot.Methods: An interpretivist qualitative design used. Semi-structured interviews five parent-child pairs were undertaken examine their during NAO...
The planning of public transport operations is an essential component urban management systems that aims to provide the most efficient, safe and effective way support movement people. Improving customer journey experience a key focus, as cities grow sustainable becomes more critical. This has led increased interest in Automatic Passenger Counting (APC) technologies real-time estimates occupancy order better information. proliferation sensors power-efficient miniaturized computing...
The study proposed a dynamic path planning (DPP) method that combines instance image segmentation and elementary matrix calculations to enable robot identify the angular position of entities in its surroundings. DPP fuses visual depth information for scene understanding estimation with reduced computation resources. This designed, developed, evaluated deep-learning based companion prototype indoor navigation obstacle avoidance using an RGB-D camera as sole input sensor. A simulated...
Abstract The transformation of cities into smarter and more efficient environments relies on proactive timely detection maintenance city-wide infrastructure, including roadside infrastructure such as road signs the cleaning illegally dumped rubbish. Currently, these tasks rely predominantly citizen reports or on-site checks by council staff. However, this approach has been shown to be time-consuming highly costly, resulting in significant delays that negatively impact communities. This paper...
With a retinal prosthesis connected to head-mounted camera, subjects can perform low vision tasks using combination of electrode discrimination and head-directed localization. The objective the present study was investigate contribution retinotopic (perception corresponding arrangement implanted electrodes with respect their position beneath retina) visual performance for three recipients 24-channel suprachoroidal implant. Proficiency in may allow good smaller head movements, identification...
The goal of this article is to provide a practical guide unlocking the power innovation within nursing community. First three intersecting components process (the itself, creativity, and environment) are explored. Next conceptual perspectives, drawn from work both von Hippel Christensen, reviewed. This followed by description methods with particular emphasis on IDEO’s innovative processes Transforming Care at Bedside initiative. Finally, specific examples organizational structures support...
We present a new visual control input from optical flow divergence enabling the design of novel, unified laws for docking and landing. While divergence-based time-to-contact estimation is well understood, use in currently assumes knowledge surface orientation, and/or egomotion. There exists no directly observable cue capable supporting approaches to surfaces arbitrary orientation under general motion. Central our measure maximum field on view sphere (max-div). prove kinematic properties...