Note the author:
Suman Raj is a Ph.D. candidate in the Department of Computational and Data Sciences at the Indian Institute of Science (IISc), Bangalore, India under the supervision of Prof. Yogesh Simmhan. Her doctoral research focuses on developing a Scalable Analysis and Planning Platform to orchestrate autonomous systems across the edge-cloud continuum. This platform is designed to support a range of mobile devices equipped with edge accelerators, allowing them to collaborate with remote servers for real-time performance in diverse, spatio-temporal autonomous applications. For instance, Unmanned Aerial Vehicles (UAVs) can be integrated into the platform to provide mobile sensing and computational capabilities for various scientific applications.
Currently, I am visiting the SciTech Group for the Fall 2024 semester, collaborating with Prof. Ewa Deelman at USC’s Information Sciences Institute on the FlyNet Project. This project, a joint effort with Michael Zink from UMass Amherst and Anirban Mandal from RENCI, initially began with a literature review focusing on escape route generation for firefighters in wildfire scenarios. After several weeks of exploration, we defined our problem statement as “Collaborative and Resilient Escape Route Planning in Wildfire Scenarios using UAVs.” This involved reviewing various open-source fire datasets, testing off-the-shelf Deep Neural Networks (DNNs) for fire detection, and training these networks to improve accuracy and performance on different edge devices, including Jetson Nano, Jetson Orin Nano, and Jetson Orin AGX.
In this project, I am fortunate to work closely with Pawel Zuk and Rajiv Mayani. Pawel has been instrumental in onboarding me to the project and assisting me with access to compute servers with the help of Mats Rynge and other lab resources. Our admin, Ciji Davis, is very approachable and helped me out with all the access that was required for me to start using the lab space. Pawel has also supported me with the system design of the project and helped me explore path planning algorithms for both drones and firefighters. Rajiv has offered valuable insights on various drone-related projects, expanding my perspective on potential applications.
A special thanks goes to Karan Vahi, who has been incredibly supportive and helped me acclimatize, when I flew to Los Angeles from India. He has also organized an engaging Pegasus tutorial, which we aim to integrate with my doctoral research work at IISc and project work at ISI. Recently, Prof. Deelman organized a seminar at ISI where I presented my research, receiving insightful feedback and questions.
Overall, this research visit has been a fantastic opportunity, providing me with valuable insights into the management of large, multi-year projects and I’m very appreciative of the SciTech Group’s guidance and collaboration. This was the first time when I traveled abroad for a long duration and it’s been a great experience with a steep learning curve.