Jebelli, graduate students receive Olympiad Medal at 2025 Olympiad in Engineering Science

Their award-winning paper, "Autonomous Drone-based System for Precision, Non-contact Surface Finishing in Construction" presents an autonomous, drone-based Unmanned Aerial Vehicle (UAV) system for high-precision, non-contact surface finishing in construction. It was selected as one of only five papers across all engineering fields to receive this distinction. 

In many modern job sites—particularly those involving elevated, confined, or sensitive surfaces—traditional manual or machine-based surface finishing methods can be inefficient, inconsistent, or even damaging. To address these limitations, Jebelli, Ren and Zhao developed a UAV platform equipped with a lightweight rotating disc and a 6-DoF gimbal system that performs hovering-based surface refinement without physical contact.  What makes this system unique is its ability to “see” and understand the surface using onboard sensors and AI, detecting subtle irregularities or imperfections and then refining those areas using a lightweight spinning tool mounted on a stabilizing robotic arm—all without making physical contact.

The system is also equipped with advanced control mechanisms to maintain steady flight and accurate positioning throughout the process through the integration of a two-layer control architecture: a low-level layer combining PID and Model Predictive Control (MPC) for flight stability, and an actor-critic reinforcement learning model to stabilize the gimbal-mounted disc against vibration and environmental disturbance. At the high level, the drone’s trajectory is dynamically guided by a transformer-based surface attention model, which fuses LiDAR and RGB data using spatial-temporal attention to identify surface zones needing refinement. Gradient-based optimization refines the coverage path in real time for high completeness and minimal redundancy.

Jebelli, Ren and Zhao used a ROS-Gazebo simulation across varying terrain conditions to validate the system, achieving over 95% surface refinement completeness, <5 mm height deviation, and sub-150 ms processing latency—demonstrating its potential for automating surface finishing with precision and minimal disturbance.

The UAV system could be applied in a number of ways in the context of construction, supporting tasks such as smoothing concrete surfaces on walls or ceilings in high-rise buildings or tunnels, polishing architectural elements like stone or finished concrete, removing uneven spray or seams from shotcrete applications, and preparing surfaces for coatings, sealants, or adhesives.

 

Beyond construction, it also has potential applications in areas such as aerospace and automotive maintenance where surface integrity is critical, ship hull finishing in hard-to-access locations, restoration of historical structures where contact must be minimized, and interior finishing tasks in hazardous or confined spaces like industrial tanks.

 

Overall, their done system enables safer, more precise, and automated surface finishing in environments where manual work would be difficult, risky, or inconsistent. It represents a step forward in developing smarter and more autonomous job sites.

 

Zhao and Ren also recently received top honors at the ASCE Global Robotics Competition with Jebelli, winning first and second place at the American Society of Civil Engineering (ASCE) International Conference on Computing in Civil Engineering (i3CE 2025) in New Orleans in May.

 

Jebelli joined the Department of Civil and Environmental Engineering at The Grainger College of Engineering in fall of 2023 as a faculty member in construction engineering management. He also serves as the director of the Robotic, Automation, and Intelligent Sensing (RAISE) Lab at the University of Illinois, Urbana-Champaign, his research centers around construction automation and robotics, particularly the integration of human physiology into robotic control and motion planning systems. 

Photo Credit: Houtan Jebelli

Photo Credit: Houtan Jebelli

Photo Credit: Houtan Jebelli