Robotics
My work encompasses robotic software and hardware development, computer vision, navigation and sensor fusion for underwater robots. As marine ecosystems change rapidly due to the effects of climate change, ocean acidification, and other human-caused pressures, we need tools to rapidly and repeatedly assess the health of these ecosystems.
My research currently focuses on vision-based Simultaneous Localization and Mapping, or SLAM, on small, low-cost portable underwater robots that can be deployed to study shallow-water environments around the world. I am working on improving the robustness of vSLAM and SFM in the marine environment by handling deformable features (e.g. soft corals, marine algae) and dynamic features.
I have deployed prototype vehicles in Hawaii and Australia to map coral reefs and am currently working on improving their capabilities to map more complex environments.
An older version of the Makobot research platform using the open-source software ORB-SLAM2 to map coral reefs and collect data for vision-based Simultaneous Localisation and Mapping development at the Hawaiʻi Institute of Marine Biology in 2018.
Vehicle field trials on Osprey Reef, Great Barrier Reef, Australia in August 2020. There were many sharks!
3D model of a section of reef at the Hawaiʻi Institute of Marine Biology. The images were collected by the Makobot robot in 2018 and processed using Agisoft Photoscan. Many thanks go to Eric Lo for processing the model and creating the fly-through of coral reefs in space.
Example of a modern SLAM system (OpenVSLAM) running in a marine environment with deforming algae features. Significant development is required to make vSLAM function robustly in these types of environments.