Summary
A team of undergraduate researchers from CMU-Q placed second in the oral presentation category at the 17th Annual UREP Awards, organized by the QRDI Council. The team developed "HyBrush," a hybrid drone system designed to clean dust from solar panels. Led by Eduardo Feo Flushing, the team utilized reinforcement learning and digital twins to train the drone to handle complex maneuvers.
DOHA, QATAR – A team of undergraduate researchers from Carnegie Mellon University in Qatar (CMU-Q) placed second in the oral presentation category at the 17th Annual Undergraduate Research Experience Program (UREP) Competition. Organized by the Qatar Research, Development, and Innovation (QRDI) Council, the prestigious competition recognizes outstanding research conducted by students across Qatar.
The team was recognized for their project titled “HyBrush: A Hybrid UAV Brushing System for PV Monitoring and Cleaning”. Led by Eduardo Feo Flushing, assistant teaching professor of Computer Science, the project addresses a challenge specific to the Gulf region: the accumulation of fine, sticky dust on solar panels which significantly reduces energy output. The team was also advised by Dhanup Somasekharan Pillai, a scientist at collaborating institute Qatar Environment and Energy Research Institute (QEERI), part of Hamad Bin Khalifa University (HBKU).
Diram Tabaa presented the project on behalf of the team at the competition.
“Most literature on solar panel cleaning drones is Eurocentric,” noted Feo Flushing. “The type of dust is different in Qatar. It’s micro-dust, more fine, and due to humidity, it gets very sticky.”
The project brought together Computer Science, Information Systems, and Computer Engineering students to develop an intelligent drone capable of flying to, landing on, and cleaning solar panels. Because unconventional drone designs are difficult to program, the team used reinforcement learning and a digital twin to train the drone in simulation, allowing it to safely learn complex maneuvers before being deployed in the real world. This approach enabled rapid experimentation with novel designs and helped bridge the gap between virtual training and real-world operation, advancing autonomous solutions for solar panel maintenance.
Feo Flushing praised the team’s resilience and the educational value of the work. “We show that there is potential to train unconventional drone designs to solve this sort of problem,” he said. “The judges valued both the message and the learning process.”
The award-winning team includes Abrar Tasneem Abir, Devang Acharya, Deep Chandra, Rama Sulaiman, and Diram Tabaa, all of whom have now completed their studies at CMU-Q. Hamad Bin Khalifa University (HBKU) student Sai Lingampalli was also on the team.
