On Wednesday the Pawsey Roadshow hit Edith Cowan University in Joondalup! It was a great turnout and Pawsey staff, ECU researchers, and attendees got together and talked high-performance computing systems and research projects.
Kieran Fazey from ECU Racing gave a presentation on the team’s project of aerodynamic design in motorsports and the methodology behind it. Kieran spoke about the pivotal role of their Computational Fluid Dynamics simulations to optimise aerodynamic design which had a significant impact on their on-track performance. The team discussed the need for Pawsey’s supercomputing infrastructure for CFD calculations which is overly time-consuming even on a high-end desktop. As a result, ECU racing may soon apply for access on our supercomputers!
Dr. Selam Ahderom, an ECU researcher, presented his project on Precision Agriculture and Throughput Computing on GPU’s. The goal is to “train” a computer vision system using machine learning algorithm (using a mathematical technique called the Deep Neural Networks) to differentiate between a weed and crop in near real time. His research group has built a prototype in their laboratory which consists of a mobile system with an inbuilt high-resolution camera. The camera takes thousands of photos of the underlying crop, and these images are mined for the presence of weeds. Graphic Processing Unit or GPU are purpose built for processing the high-resolution image data and the application of machine learning techniques. Dr. Ahderom has been using the world-class facilities at Pawsey Supercomputing to store, process and share the vast amounts of data needed for the project. Precise sensing of weed presence would mean targeted application of herbicides. Dr. Ahderom’s goal is to install this weed detecting system onto farm machinery which otherwise blanket-spray crops with herbicide. The researcher hopes that this technology will result in the reduction of herbicide use in crops significantly.
Go to our Facebook page to see photos of the event.