Yogesh Girdhar

YorkU’s EECS Seminar Series (YESS)

October 24, 2024 2:00 pm EST at Lassonde Building, room LAS-3033

Coral reef ecosystem monitoring using autonomous robots

Evaluating coral reef biodiversity and ecosystem health is currently an extremely time consuming and expensive operation, requiring divers and trained marine biologists. This creates a monitoring bottleneck limiting timely detection of ecosystem changes due to natural or anthropogenic causes, and as a result limiting potential interventions that could be deployed to reverse these negative changes.

We present a novel system for scaling up coral reef ecosystem monitoring. The proposed system consists of a fully autonomous robot – CUREE and a web based survey processing pipeline – tektite.ai. CUREE AUV is equipped with multiple cameras, hydrophones and computing power to enable adaptive visual and passive acoustic guided surveys. It is capable of automatically discovering bioactivity hotspots using visual and passive acoustic cues; autonomously conduct low altitude surveys for high resolution benthic imaging, and visually track arbitrary animals to observe their behavior. Observation data from CUREE is processed by the tektite.ai web service that can produce high resolution 3D maps, run neural networks to quantify coral cover, animal abundance, rugosity, and other statistics. Images collected by CUREE are corrected for distortions due to backscatter and attenuation using the DeepSeeColor algorithm to produce accurate representation of color in the 3D model. Results of analysis using various neural nets are visualized in 3D to enable reef managers and stakeholders to get the most immersive understanding of the reef ecosystem state. Ongoing efforts are focused on improving the performance of the various neural networks, and to make the robot deployable by non-experts. The proposed approach aims to be the most rapid, cost effective, and comprehensive way to monitor the state of coral reef ecosystems.

Bio

Yogesh Girdhar is a computer scientist, and the PI of the WARPLab (http://warp.whoi.edu), and Associate Scientist (with tenure) at Woods Hole Oceanographic Institution (WHOI), in the Applied Ocean Physics & Engineering department. He received his BS and MS from Rensselaer Polytechnic Institute in Troy, NY; and his Ph.D. from McGill University in Montreal, Canada. During his Ph.D. Girdhar developed an interest in ocean exploration using autonomous underwater vehicles, which motivated him to come to WHOI, initially as a postdoc, and then later continue as a scientist to start WARPLab. Girdhar’s research has since then focused on developing autonomous exploration robots that, through the use of AI, can accelerate the scientific discovery in the oceans. Some concrete examples of his research interests include: underwater 3D scene reconstruction with accurate physics based appearance modeling; autonomous marine animal behavior monitoring using visually guided robots; unsupervised substrate characterization; interactive vision guided underwater exploration over low-bandwidth communication channels; detecting marine bioactivity hotspots using passively listening AUVs; learning to swim acrobatically using reinforcement learning; and generative models for distribution of high dimensional categorial observations such as phytoplankton taxa. Some notable recognition of his work includes the Best Paper Award in Service Robotics at ICRA 2020, finalist for Best Paper Award at IROS 2018, and honorable mention for 2014 CIPPRS Doctoral Dissertation Award.

EECS Upcoming Events: https://lassonde.yorku.ca/eecs/eecs-upcoming-events/