117 research outputs found
Implementation of insect vision based motion detection models using a video camera
No full textDespite their limited information processing capabilities, insects (with brains smaller than a pinhead) are able to manoeuvre with precision through environments that are highly-crowded and contain moving objects. Their ability to avoid collisions using limited computing power forms the basis for this project, in which we attempt to simulate the motion detection ability of insects using two models - the Horridge Template Model and the Reichardt Correlation Model. In this project, the direction of motion of a moving object and its angular speed are determined by capturing visual data using a web camera focussed on a moving pattern generated by VisionEgg software. The performance of both the models is quantitatively compared and various error-reducing techniques are investigated.Andrew Budimir, Sean Correll, Sreeja Rajesh, and Derek Abbot
Corrigendum: Network analysis of the relationship between depressive symptoms, demographics, nutrition, quality of life and medical condition factors in the Osteoarthritis Initiative database cohort of elderly North-American adults with or at risk for osteoarthritis (Journal of Physical Chemistry (2019) DOI: 10.1017/S204579601800077X)
No full textThe above article was originally submitted to Epidemiology and Psychiatric Sciences with the incorrect author name listed for the second author. The correct surname for the second author is Ai Koyanagi (and not Ai Konayagi). The authors would like to apologise for this error. © Cambridge University Press 2019
Corrigendum: Network analysis of the relationship between depressive symptoms, demographics, nutrition, quality of life and medical condition factors in the Osteoarthritis Initiative database cohort of elderly North-American adults with or at risk for osteoarthritis (Journal of Physical Chemistry (2019) DOI: 10.1017/S204579601800077X)
No full textThe above article was originally submitted to Epidemiology and Psychiatric Sciences with the incorrect author name listed for the second author. The correct surname for the second author is Ai Koyanagi (and not Ai Konayagi). The authors would like to apologise for this error
MAGPIE2: Designing a Robust and Manufacturable Robot Gripper with Built-in 3D Perception, Speed, Force, and Position Control for Autonomous Manipulation
No full textThis paper presents the design and development of a versatile robot gripper, the MAGPIE2, that prioritizes strength, robustness, ease of manufacturing, and eye-in-hand capabilities. Unlike grippers tailored for specific tasks or objects, this gripper aims for generality through a linkage design. The entire gripper is designed using cheap materials like 3D-printed PLA and inexpensive fasteners, employing a simple 4-bar linkage system. Motors with speed, position, and force control capabilities are selected to drive the gripper. The developed gripper exhibits 32 Newtons of force at the fingertips and achieves tasks with tolerances below <0.5mm. Extensive testing on the widely used YCB dataset demonstrates the gripper's versatility with a success rate of 93\%. Moreover, 3D perception integrated at the palm of the hand enhances visual feedback throughout the grasping process. The assembly of the Siemens gear assembly problem, a sensor-based stacking, and force control experiments are conducted to evaluate the gripper's precision, strength, force control, and camera capabilities.</p
MAGPIE2: Designing a Robust and Manufacturable Robot Gripper with Built-in 3D Perception, Speed, Force, and Position Control for Autonomous Manipulation
No full textThis paper presents the design and development of a versatile robot gripper, the MAGPIE2, that prioritizes strength, robustness, ease of manufacturing, and eye-in-hand capabilities. Unlike grippers tailored for specific tasks or objects, this gripper aims for generality through a linkage design. The entire gripper is designed using cheap materials like 3D-printed PLA and inexpensive fasteners, employing a simple 4-bar linkage system. Motors with speed, position, and force control capabilities are selected to drive the gripper. The developed gripper exhibits 32 Newtons of force at the fingertips and achieves tasks with tolerances below <0.5mm. Extensive testing on the widely used YCB dataset demonstrates the gripper's versatility with a success rate of 93\%. Moreover, 3D perception integrated at the palm of the hand enhances visual feedback throughout the grasping process. The assembly of the Siemens gear assembly problem, a sensor-based stacking, and force control experiments are conducted to evaluate the gripper's precision, strength, force control, and camera capabilities.</p
The collapsible space between us : the interrelationship between testifier, author, and reader in Dave Eggers's "What Is the What"
No full textThesis (M.A.)--Georgetown University, 2009.; Includes bibliographical
references. This project investigates the collaborative relationship between testifier
(Valentino Achak Deng, a Lost Boy of Sudan), author (Dave Eggers), and readers in Dave Egger's
What Is the What. I explore the changing genre of memoirs. I use narrative and reader-response
theories to analyze Eggers's meticulous narrative construction. Finally, I argue that Eggers
builds a collaborative relationship with the reader in order to transform them into an
activist outside of the text
ULF induction magnetometer data from Australian Antarctic and subantarctic stations
No full textProgress Code: onGoingStatement:
It is impossible to name all the expeditioners who have assisted with collecting these data. In each case they are the respective Physicists at each base, or if there is no such physicist, the electronic engineer. In recent years there have been no such person at some locations, e.g. Macquarie Island, and the BoM technician has been helpful (e.g. Michael Hyde). Again, it would be necessary to conduct extensive trawls through HR records from a range of institutions to get this information.
At AAD the main people involved with collecting the data as it arrives from the bases, passing it to our server at Newcastle, and providing it to the international space weather centre at the Bureau of Meteorology are Damian Murphy and Lloyd Symons. They have been very helpful and supportive over many years. Recently David Correll has also been involved with this process. Broad direction has been provided by Andrew Klekociuk.
At the University of Newcastle the contact people are:
Prof Fred Menk, [email protected]; Prof Colin Waters, [email protected];
Dr Sean Ables, [email protected]
The magnetometer at Macquarie Island has ceased functioning and so no data are coming from there. Otherwise the data collection and transfer process works seamlessly.Space weather is a significant hazard to modern technology. It results from energization of the Earth's radiation belts and subsequent transport of relativistic particles to low altitudes. Plasma waves convey energy and interact with particles throughout these regions but also provide a convenient ground-based diagnostic of these processes. We will use data from current and planned spacecraft missions referenced to ground observations to determine the sources of these waves, and whether they are responsible for acceleration of particles in geospace and their precipitation to the atmosphere.<br/><br/>Induction magnetometers constructed by the University of Newcastle and AAD have been installed and operating at each base for over 2 decades. Some people refer to such instruments as search coil magnetometers. Time series observations for the two horizontal magnetic components are digitally recorded with cadence of at least 1 Hz. The data are telemetered from the bases to HO Kingston from where the data are loaded to an automatic FTP site for transfer to Newcastle and the IPS Australian Space Weather Centre at the Bureau of Meteorology. At Newcastle University the data are stored on servers and are automatically backed up by research IT support services. The data transfer process was established by Lloyd Symons and has operated with no difficulty for many years. The main parties involved are Waters, Menk, Ables and Fraser at Newcastle and Lloyd Symons and Damian Murphy at AAD. The induction magnetometer data are archived and freely available through the Bureau of Meteorology's World Data Centre and the AAD can also archive and use these data as desired.<br/><br/>The induction magnetometer data can be viewed in near-real time at the IPS web site; see http://www.ips.gov.au/Geophysical. Since these are induction data, the signals are dB/dt and also have a frequency-dependent phase shift. The spectral slope of 1/f compensates for the natural response of the geomagnetic field, and the frequency range over which geomagnetic variation signals are usefully detected is 1 mHz to 3 Hz. The signal amplitude is in relative units. Contact Newcastle for calibration information regarding absolute units and phase responses
Improving Autonomy for Precision Restoration of Degraded Rangelands
No full textRangelands cover approximately 41% [41] of the Earth’s land surface and support about 38% of the global population. However, 20% [2] of these rangelands are considered degraded due to poor land management, climate change, and increased development. Restoring these degraded areas is paramount for long-term sustainability and requires intervention which is intractable for humans to perform at scale due to the sheer volume of area. Autonomous robots present a promising alternative but despite advancements in autonomy and machine learning, there are still several challenges that need to be addressed if we are to develop a platform suitable for making meaningful interventions in these environments.
Over the course of two years, we deployed to a degraded rangeland four times, giving us a unique insight into the unique challenges posed by the autonomous revegetation problem. We identify long-term data association with sparse observation as the key problem facing autonomous revegetation. This problem is fundamental to making impactful intervention as tracking the outcomes of any intervention at a high resolution is prerequisite to understanding their efficacy. Longterm data association of vegetation in these environments is complicated by three sub-problems; localization in agricultural UGVs, image segmentation for vegetation, and visual distinction.
Furthermore, vegetation violates the static-world assumption that underlies classic mapping methods. In order to accurately localize the vegetation (which is a key part of tracking and uniquely identifying them), one must address the semi-static nature of vegetation as a landmark. Seasonal changes can confuse data association. In my dissertation I will provide a statistical, feature-based method for anticipating these changes and leveraging pre-existing knowledge of these kinds of changes for long-term mapping in the agricultural context.</p
Navigating Visually Degraded Environments Using Millimeter Wave Radar
No full textTechniques using millimeter-wave radar to perform a variety of metric and semantic robotics perception tasks have increased greatly in the past decade (Harlow 2024). The decrease in cost, and unique ability of radar sensors to see through visually degraded environments marks the sensor as a compelling alternative to visual and near visual spectrum sensors such as cameras and lidar. Radar-based navigation still lacks in some core areas, however. Specifically, there are no dense front ends using radar images for odometry, partially due to fundamental difficulties in using this data. This work presents an investigation into using dense radar images as a standalone odometry sensor as well as fusion with other common robotics sensors for use in odometry applications.</p
Navigating Visually Degraded Environments Using Millimeter Wave Radar
No full textTechniques using millimeter-wave radar to perform a variety of metric and semantic robotics perception tasks have increased greatly in the past decade (Harlow 2024). The decrease in cost, and unique ability of radar sensors to see through visually degraded environments marks the sensor as a compelling alternative to visual and near visual spectrum sensors such as cameras and lidar. Radar-based navigation still lacks in some core areas, however. Specifically, there are no dense front ends using radar images for odometry, partially due to fundamental difficulties in using this data. This work presents an investigation into using dense radar images as a standalone odometry sensor as well as fusion with other common robotics sensors for use in odometry applications.</p
- …
