Talent: Research & Innovation 🔬
Fulton Undergraduate Research Initiative (FURI)
Fall 2024, Spring 2025, Summer 2025 | Faculty Mentor: Dr. Paul Grogan
As part of FURI in Fall 2024, I started pursuing research under Dr. Paul Grogan on Reinforcement Learning Techniques for autonomous satellite orientation in agile space missions for Earth Observation Systems. Dynamic Targeting is an extensively researched concept used by satellites to point the instruments in the optimal direction to get the maximum scientific yield.
Research Poster & Presentation
Research poster presented at FURI Symposium showcasing methodology and results
The Challenge
Conventional methods use scheduling systems for determining points on the map that are worth observing. Auto-scheduler systems are effective in selecting priority observations but these methods are not real-time or fully accurate. Incorrect decisions by satellites lead to wastage of power and redundant data. This research aims to enable satellites to autonomously determine valid observation locations in real time during Earth Science Missions.
Research Approach
The project focuses on creating an RL environment that emulates the satellite's decisions to measure over the priority observational points. This involves:
- Usage of large simulated environmental data in NASA GEOS-5 dataset
- Model selection and training
- Testing and tuning of the model
- Evaluation through comparison with a random forest classifier model
Current Results
Ongoing research has seen a policy-based DQN model and a Quantile-Regression DQN model perform with 79% recall. However, the results showed a low F1 score due to more false positive actions. The low precision is potentially due to oversampling near the 0° latitude and the huge class imbalance in the datasets derived from NASA GEOS-5.
Future Work
The project focuses on preparing larger training datasets of 8-12 months with additional features like cloud optical thickness and incoming shortwave flux. The project further aims to use Decision Transformers or STTNs in ensemble with reinforcement learning to tackle the oversampling challenge.
The research will be considered a success when the results correspond with the simulated data on convective precipitation storms, proving a greater precision than predictive supervised learning models like the random forest classifier.
Impact & Skills
The ultimate goal of the research is to autonomously enable satellites in prioritizing areas with high atmospheric activity, enhancing environmental monitoring and providing deeper insights to Earth's atmospheric dynamics. This aligns with the theme "Joy of Living" as it leads to more useful data yielded that affect a variety of applications like:
Meteorological predictions
Disaster relief
Geological surveys
Defense strategies
This research also made me competent in industrial skills like Python, handling big data servers like OpenDAP, and heavy mathematical concepts behind deep reinforcement learning.
Connection to "Joy of Living" & Personal Impact
GCSP Theme Relevance
This research directly contributes to the "Joy of Living" theme by improving the quality and accuracy of Earth observation data. Better satellite targeting means more effective disaster preparedness and response, saving lives during hurricanes, floods, and severe storms. Enhanced meteorological predictions help farmers optimize crop yields, coastal communities prepare for weather events, and cities better manage resources. By making satellite systems smarter and more autonomous, we enable better decision-making that protects communities and improves daily life for millions of people worldwide.
Professional Development
This research experience has been transformative for my career trajectory in aerospace engineering and machine learning. Working with cutting-edge reinforcement learning techniques and NASA datasets has given me hands-on experience with technologies used by leading space agencies and tech companies. I've developed proficiency in Python, PyTorch, handling large-scale environmental data through OpenDAP servers, and implementing complex deep learning architectures. The experience of presenting at the FURI Symposium has improved my scientific communication skills, while regular meetings with Dr. Grogan have taught me how to approach research methodically and think critically about results. These skills are directly applicable to careers in autonomous systems, satellite engineering, and AI research.
Personal Growth
Beyond technical skills, this research has deepened my appreciation for how engineering can address real-world challenges. Working through setbacks—like the class imbalance issues and low precision scores—has taught me resilience and the importance of iterative problem-solving. I've learned that research is not just about achieving perfect results, but about understanding why things work or don't work, and persistently refining approaches. This experience has reinforced my passion for using technology to make a tangible positive impact on society, particularly in environmental monitoring and climate science. It has also shown me the joy that comes from contributing to something larger than myself—work that will benefit communities around the world through better disaster preparedness and resource management.
Funding: FURI and GCSP Research funding ($4,600 total)