Skip to content. Skip to navigation
CIM Menus

Challenges in Autonomous Robotic Exploration and Monitoring

Yogesh Girdhar, Postdoctoral Fellow
Applied Ocean Physics & Engineering Woods Hole Oceanographic Institution

March 19, 2015 at  10:00 AM
McConnell Engineering Room 437


Autonomous exploration robots can be used for monitoring tasks that might be too expensive or too dangerous for humans to perform. We want to explore distant planets, monitor our forests for fires, diseases and illegal logging, our borders and coastlines for anomalies, and our oceans for archeological discoveries and evaluating animal biodiversity. In this talk I will present my ongoing research on some of the issues that arise with the use of such autonomous exploration robots.

First I will describe a novel topic modeling based approach to modeling curiosity in a mobile robot, which is useful for monitoring and adaptive data collection tasks. We have validated the technique using simulated exploration experiments using aerial and underwater data, and demonstrated an implementation on the Aqua underwater robot in a variety of scenarios. Second, I will describe techniques to summarize the data collected by a robot on an exploration mission, online, and in batch. The proposed navigation summaries aim to select a small subset of the collected data that is representative of the diversity of what was observed, and hence is useful for providing live mission updates over a limited bandwidth communication channel. Third, I will describe the irreversible sample collection problem, which can be posed as an instance of the classical secretary hiring problem. I will present solutions to a few variants that are optimal under different conditions. Finally, I will present a brief summary of my ongoing work on evaluating marine habitats using autonomous robots and unsupervised machine learning techniques, and discuss some ideas for future work.


Dr. Yogesh Girdhar is Devonshire Postdoctoral Scholar, and FQRTN Postdoctoral Fellow at Woods Hole Oceanographic Institution, in the Applied Ocean Physics & Engineering department. He received his BS and MS from Rensselaer Polytechnic Institute in 2005, where he was awarded Paul A. McGloin prize for most outstanding academic achievement in Computer Science. He received his PhD from McGill University in 2014 with thesis title “Unsupervised Semantic Perception, Summarization, and Autonomous Exploration for Robots in Unstructured Environments”, which received the Honorable Mention for CIPPR Doctoral Dissertation Award. Yogesh’s research interests are in the algorithmic challenges that lie on the intersection of field robotics, machine learning and computer vision.