Team Leader, Robotics, IEEE Senior Member, A/Editor IEEE RA-L
Research Leader, Visual-AI, Intelligent Automation (IA) Group
Multi-Sensor Fusion & Machine Vision
Multi-Sensor Fusion & Machine Vision
AI and Machine Learning
Energy and Machine Learning
HDR Candidate
Data-efficient Machine Learning for Robotic Perception and Manipulation
HDR Candidate
Extended Robotic Reality for Robot Learning
HDR Candidate
Robotic Intelligence for Construction Waste Sorting
HDR Candidate
Vision Systems for Processes Control in Metal Additive Manufacturing
HDR Candidate
Developing the Next Generation Materials Science Lab
HDR Candidate
Additive Manufacturing
Human-Centered Automation
Aerospace & Advanced Manufacturing
Manufacturing & Materials
Deploying robots in collaboration with humans is seen as an enabler of major changes in construction productivity for various tasks, such as digital twain, quality/compliance inspection, progress monitoring and automated interior finishing.
IEEE Transactions on Industrial Electronics, 2020 IROS 2020 Mechanism and Machine Theory Robotics and Computer-Integrated Manufacturing FToMM Symposium on Robot Design, Dynamics and Control
XR devices offer a range of spatial perception capabilities that can enhance human and robot collaboration. From 6DoF tracking to depth perception, gesture recognition, and environmental mapping, these features enable applications spanning human and robot interaction for manufacturing, healthcare, and beyond, XR devices continue to push the boundaries of spatial perception and redefine how we interact with robots
This is a generic article you can use for adding article content / subjects on your website.
Robotic additive manufacturing is a technology that can deposit material and fabricate complex parts. Part geometry during the process can negatively affect the shape and size of the final manufactured object. In-process spatio-temporal 3D reconstruction, also known as 4D reconstruction, allows for early detection of deviations from the design in robotic additive manufacturing, thus providing the opportunity to rectify at an early stage, making the process more robust, efficient and productive.
In our study, aerial images from the suburbs of Melbourne revealed two primary types of soiling on PV panels: dust, attributable to wind and climate conditions, and bird droppings, distinct from vegetation debris like leaves or shading effects. Soiling on solar panels directly impacts their efficiency, with bird droppings, despite their small size, having a substantial effect on overall output. The size and specific location of these droppings are critical factors in determining the extent of power loss. Over time, bird droppings can lead to more severe, long-term issues, including permanent panel damage, primarily due to hotspots created by their accumulation. These soiling types require different cleaning approaches; dust can be removed with low-pressure water jets, whereas bird droppings require a specialized cleaning solution for effective removal. Consequently, our dataset was curated to reflect these cleaning requirements. It was observed that the ratio of dust to bird droppings on soiled PV panels averages 1:2, leading to a class imbalance challenge in the dataset. Further complicating detection, bird droppings are small, lack distinct spatial or color features, and generally cover less than 2% of a panel's surface, contrasting with larger dust patches. The panels, as detailed in Table , are of the Polycrystalline type with a blue background, where bird droppings typically appear white or grayish, without a specific shape. These factors collectively contribute to the complexity of accurately detecting soiling on PV panels.