Clark Abou-Rjeily

Let's build robots!

Georgia Institute of Technology

I specialize in Controls, AI, and Perception.
I build intelligent autonomous systems through multi-robot coordination, deep reinforcement learning, and vision-based control.
I'm currently completing my Master's degree at Georgia Tech.

</developer>

Get In Touch View Projects

About Me

My name is Clark, and I am an electrical, computer, and robotics engineer with a strong interest in building intelligent systems that can reason, adapt, and operate reliably in complex real-world environments. I am currently completing my Master of Science in Robotics at Georgia Institute of Technology, where I focus on bridging theory with hands-on system development. I'm excited about what lies ahead!

My technical interests sit at the intersection of controls, artificial intelligence, and perception. Through my work with Georgia Tech’s FACTS Lab, I have explored problems in multi-robot coordination and autonomy, with an emphasis on designing systems that are robust, scalable, and grounded in real-world constraints.

I earned my undergraduate degree in Electrical and Computer Engineering from the American University of Beirut, where I built a strong foundation in control systems, machine learning, and instrumentation through research-driven projects and engineering-focused coursework.

Outside of robotics, I am a Level VIII certified piano player. Years of musical training taught me discipline and attention to detail. Also, I love skiing, and most of my vacation days will align with good powder days!

Technical Projects

A selection of my work in autonomous systems, robotics perception, and multi-robot coordination.

Current Work

LUCID: Look-ahead Unified Camera-LiDAR Inference and Diffusion

LUCID is a deep learning project for autonomous driving that predicts future LiDAR readings from current multimodal sensor inputs (LiDAR, camera, radar, and vehicle state). By generating look-ahead sensor data, the system enables vehicles to anticipate their surroundings earlier; improving reaction time, long-horizon planning, and overall safety. TL;DR: Multimodal DiT pipeline for dynamics-conditioned future 3D LiDAR generation on nuScenes data.

Autonomous DrivingDynamicsLiDARDiffusionTransformerPython

LLM-Guided Heterogeneous Swarm Navigation & Negotiation

This capstone studies how humans can specify complex heterogeneous multi-robot objectives in natural language and execute them reliably in dynamic environments. High-level intent is translated into formal task constraints, enabling decentralized planning, execution, and real-time monitoring of swarm behavior. When missions become infeasible due to environmental changes or conflicting constraints, an LLM-driven negotiation layer proposes minimal task revisions while preserving the original intent. TL;DR: Natural language tasking → formal constraints → swarm planning → LLM/VLM-based task repair.

Swarm RoboticsNavigationTask PlanningLLM/VLMMulti-AgentTemporal Logic

Dual-Arm Industrial Manipulation

Ground up implementation of bimanual coordination for ROS-based tasks using advanced kinematics and force control, using DENSO VM-60B1 arms.

ManipulationBimanualKinematicsROS 2CADSolidWorksBlenderGazeboRViz

Previous Work

Decentralized Heterogeneous ACO

Ant Colony Optimization framework for capability-based multi-agent cooperation on Georgia Tech's Robotarium integrated with Reinforcement Learning and reward shaping, adapted for lower communication ranges. Customized control barrier functions are used for collision avoidance through quadratic programming. RL results in emergent per-agent role specialization, i.e. exploration vs. exploitation.

Multi-AgentACORobotariumQPOnline RL

DIAYN with Human-Relevant Skills

Reproduction and validation of Diversity Is All You Need (DIAYN) across benchmarks, extended to 6-DOF Panda Arm Manipulation in robosuite with HRL/VAE. The extensions were made in order to contextualize and guide the skill-learning to achieve relevant and more useful skills.

Deep RLPanda ArmSACVAEPyTorchMuJoCo

Multi-Robot Conflict Resolution

Multi-agent planning in constrained environments, focusing on bottleneck management and safe coordination. Coalition-aware abstraction to improve synchronization and eliminate inter-robot collisions.

Multi-AgentCollision AvoidanceNavigationMAPF

HRI Emotion-Conveying Robot Sounds

Study on how non-verbal robot sounds can convey emotion to improve human-robot interaction. The study evaluated the performance of our emotional sounds-generating diffusion model against different modalities.

HRIUser StudyDiffusionAudio

The Robot Collective

Traffic-Aware Multi-Agent Path Planning with dynamic edge weighing for block-stacking bots in ROS. Emergent learned traffic flow over time.

ROSPath PlanningA*Multi-AgentGraph TheoryControl

Mixed Reality Mobile Manipulator Virtual Testbed

Unity3D AR application for human-robot collaboration via ROS 2 and MoveIt using a HoloLens2. We used a virtual UR5e arm mounted on a real Clearpath Husky.

Unity3DARHoloLensROS 2HRC

Sensor Fusion for 3D Reconstruction

3D scene reconstruction from vision data for robotics and SLAM applications, used to generate a novel multi-sensor 3D reconstruction dataset with the use of motion capture.

3D ReconPerceptionMocapOptiTrackC#

Resilient Quadrotor Control

Implementation of LQR and other optimal control strategies for quadrotor stability, disturbance rejection, and trajectory tracking.

LQROptimal ControlQuadrotorsSimulink

Instrumentation Smart Bicycle

IoT-enabled bicycle with sensors for performance monitoring and rider safety, using an Arduino Uno.

InstrumentationArduinoIoTSensorsCAD

Education

Georgia Institute of Technology

M.S. in Robotics

08/2024 – 05/2026

Atlanta, GA

Core Areas: Controls, Artificial Intelligence (AI), and Perception
GPA: 4.0/4.0 | Graduating Spring 2026
Coursework: Robotics, Nonlinear & Networked Control, Deep Reinforcement Learning, Computer Vision, Human-Robot Interaction

American University of Beirut (AUB)

B.Eng. in Electrical and Computer Engineering

08/2020 – 06/2024

Beirut, Lebanon

Minor in Business Administration
GPA: 4.0/4.0 | Graduated with High Distinction, ranked 1st
Track in Control, Robotics, and Intelligence Systems
Coursework: Autonomous Mobile Robotics, Machine Learning, Modern Control Systems, Instrumentation

Experience

FACTS Lab, Georgia Tech

Research Assistant

05/2025 – Present

Atlanta, GA

Integrating 6-DOF industrial manipulators for ROS2-based multi-agent collaboration and VLM-driven tasks
Enabling safety-critical bimanual planning and control with Signal Temporal Logic (STL) constraints

STAR Lab, Georgia Tech

Research Assistant

01/2025 – 04/2025

Atlanta, GA

Developed a motion planner using nonlinear optimization for heterogeneous multi-robot coordination
Added a coalition-aware abstraction to improve synchronization control and eliminate inter-robot collisions

Vision and Robotics Lab (VRL), AUB

Research Intern

06/2023 – 08/2023

Beirut, Lebanon

Generated a novel multi-sensor, asynchronous, real, and collaborative 3D reconstruction dataset
Conducted motion capture, experiment design, and sensor fusion using C#, OptiTrack Motive, Cyclone 3DR
Co-authored a research paper for IROS, 2nd author, Didymos-XR Project

American University of Beirut (AUB)

Teaching Assistant

09/2022 – 05/2024

Beirut, Lebanon

Cooperated with professors in teaching Electronics I & II to junior ECE students
Enhanced concept delivery and graded weekly assignments for 50+ students

L’Art de la Musique (School of Music)

Teacher

09/2022 – 09/2023

Beirut, Lebanon

Conducted Piano, Music Theory, and Aural Skills courses to students of ages 6 to 30+
Certified Level VIII Piano Performance by the Lebanese National Higher Conservatory

Professional Tutoring

Self-Employed

04/2019 – 06/2024

Beirut, Lebanon

Held sessions for Baccalaureate and Electrical Engineering students, improving grades by multiple letters
Focused on Mathematics, Physics, and Core Engineering subjects

Build It Weekend 5.0 (AUBRC x IEEE Robotics Competition)

Participant (2nd Place)

03/2023 – 04/2023

Beirut, Lebanon

Won 2nd place in a 48-hour robotics competition out of 10 teams, judged by AUB professors
Designed and built a prototype lifeline rescue-bot with IoT control using ESP32