Academic Website • Robotics • Autonomous Systems

Mohamed Elgouhary

Ph.D. Student • Robotics • Autonomous Driving • Reinforcement Learning

I am a Ph.D. student and Graduate Research Assistant at West Virginia University working on autonomous driving, autonomous racing, LiDAR-based perception, learning-enabled control, and sim-to-real robotic systems.

Mohamed Elgouhary

Quick Info

Location: Morgantown, WV, USA

Affiliation: Lane Department of Computer Science and Electrical Engineering, West Virginia University

Research: Safe autonomous systems, learning-enabled control, LiDAR perception, embodied AI

About

My research focuses on autonomous driving and autonomous racing, with emphasis on real-time control, robust perception, and reinforcement learning for safety-critical decision-making. I develop ROS2-based autonomy stacks that combine global path tracking, local obstacle avoidance, LiDAR perception, and safety mechanisms for real-world robotic systems.

My current interests include safe autonomous systems, sim-to-real transfer, learning-enabled control, embodied AI, and foundation-model approaches for high-level robotic reasoning.

Professional Experience

West Virginia University — Ph.D. Student & Graduate Research Assistant

Morgantown, WV, USA • Aug. 2024 – Present

Conducting research in autonomous driving and racing using ROS2, LiDAR-based perception, reinforcement learning, and sim-to-real deployment on Roboracer platforms.

High Institute of Engineering and Technology — Teaching Assistant

Egypt • Sept. 2013 – May 2024

Supported teaching in mechatronics and engineering courses, supervised student projects, and contributed to academic instruction in programming, mathematics, and systems topics.

Jelecom for Technical Training Company — Software Engineer

Cairo, Egypt • Sept. 2012 – Sept. 2013

Worked in software engineering and technical training, supporting applied computing workflows.

Selected Projects

Representative work in learning-enabled control, racing autonomy, and sim-to-real robotic systems.

RL-Based Tuning of Pure Pursuit for Autonomous Racing

Developed a reinforcement learning framework to jointly tune lookahead distance and steering gain in Pure Pursuit, with sim-to-real validation on Roboracer platforms.

Safety-Gated Arbitration for Global–Local Control

Designed a learning-enabled arbitration framework that combines global raceline tracking with local obstacle avoidance under safety constraints.

Sim-to-Real Roboracer Deployment with LiDAR Perception

Built ROS2-based autonomous racing systems integrating LiDAR perception, Pure Pursuit, Gap-Follow, safety bubbles, and speed limiting for physical Roboracer vehicles.

Embodied AI and Foundation-Model Exploration

Exploring how LLMs, VLMs, and related foundation-model approaches can support high-level reasoning and decision-making in safety-critical autonomous systems.

Publications

Published journal articles, accepted workshop papers, and current manuscripts.

Published Journal Articles

Learning to Tune Pure Pursuit in Autonomous Racing: Joint Lookahead and Steering-Gain Control With PPO
IEEE Robotics and Automation Letters, 2026

Journal article on reinforcement learning for adaptive Pure Pursuit in autonomous racing, with emphasis on controller tuning, racing performance, and sim-to-real deployment.

A proposed video super-resolution reconstruction strategy using wavelet multi-scale convolutional neural networks
Journal of Optics, Springer, 2023

Research on video super-resolution using wavelet multi-scale convolutional neural networks for improved reconstruction quality.

Accepted Workshop Papers

Dynamic Lookahead Distance via Reinforcement Learning-Based Pure Pursuit for Autonomous Racing
Autoware Workshop, IEEE Intelligent Vehicles Symposium (IV 2026)

Workshop paper on learning a dynamic lookahead policy for Pure Pursuit in autonomous racing, with relevance to adaptive control and sim-to-real deployment.

Robust Global–Local Behavior Arbitration via Continuous Command Fusion Under LiDAR Errors
Autoware Workshop, IEEE Intelligent Vehicles Symposium (IV 2026)

Workshop paper on robust and safety-aware arbitration between global path tracking and local obstacle avoidance under LiDAR sensing errors.

Education

West Virginia University

Ph.D. in progress • Aug. 2024 – Present

Lane Department of Computer Science and Electrical Engineering.

Mansoura University

M.Sc. in Electronics and Communications Engineering • Sept. 2017 – Feb. 2023

Thesis: Multi-Frame Super-Resolution using Wavelet Multi-Scale Convolutional Neural Networks.

Mansoura University

B.Sc. in Electronics and Communications Engineering • Sept. 2007 – Jun. 2012

Graduation Project: Wireless Brain-Controlled Humanoid System.

Technical Skills

Core tools and topics used across my research, development, and deployment workflows.

ROS2 Python C/C++ LiDAR Perception Pure Pursuit Gap-Follow Reinforcement Learning PPO F1TENTH Gym Git/GitHub Linux Autonomous Racing

Teaching & Outreach

Teaching Experience

  • Modelling and Simulation
  • Algebra and Differential Calculus
  • Analytical Geometry and Integral Calculus
  • Ordinary Differential Equations
  • Partial Differential Equations
  • Computer Programming

Outreach

Arabic-language AI content creator on YouTube since 2020, producing educational content in artificial intelligence, machine learning, mathematics, and programming.

Contact

Mohamed Elgouhary

Email: mohamed.elgohary.ai@gmail.com

Academic Email: mae00018@mix.wvu.edu

Location: Morgantown, WV, USA

Affiliation: Lane Department of Computer Science and Electrical Engineering, West Virginia University