Current Research
DRL-Based Pathfinding, For Multi-Robot Systems
Traffic Aware, Multi-Agent PathFinding (MAPF)
Using Deep Reinforcement Learning (DRL) to optimize Multi-Agent Pathfinding (MAPF) solutions.
Distributing learned policies to agents, leveraging local observations and implicit inter-agent communication for action determination.
scalable up to 1000 robots in the de520d map from the MAPF benchmark.
Robust against probabilistic delays in robot movements.
Centralized learning with a decentralized execution model.
Developed in Python, utilizing the PyTorch framework.
Employs predictive traffic density analysis to optimize path planning, resulting in fewer conflicts.
Implements a decentralized execution model for enhanced autonomy and scalability.
Multi-UAV 4D Trajectory planning.
Each agent reserves the airspace using 4D boxes for a safer flight
Effecient planning using Stochastic based descritization
About Me
I'm a passionate and versatile postdoctoral researcher at the Southern Denmark University (SDU) with a focus on robotics My journey in academia has been marked by innovation, collaboration, and a commitment to excellence
Portfolio
At SDU, my work is focused on, developing new approaches for multi-robot coordination, enabling fleets of robots to safely and efficiently navigate shared environments. This experience reflects my dedication to delivering solutions that not only push technological boundaries but are also efficient and accessible.
Beyond my core projects, I actively publish robotics coding tutorials on my YouTube channel, firmly believing in the power of knowledge sharing through small-scale projects. It's my way of giving back to the tech community and getting broader feedback on my work.
Publications
M. Boumediene, L. Mehennaoui, and A. Lachouri, FDA*: A Focused Single-Query Grid-Based Path
Planning Algorithm,” Journal of Automation, Mobile Robotics and Intelligent Systems (JAMRIS), pp. 37–43,
May 2022.
M. Boumediene, N. Zeghida, B. Manaa, and D. L. Mehennaoui,” Design, Construction, and Control of a
Self-Balancing Robot Including a New Frame Assembly Approach and a Custom PCB ”, in Proceedings of the
International Conference on Technological Advances in Electrical Engineering (ICTAEE), Skikda, Algeria,
May 2023, pp. 200–206.
M. Boumediene, S. Laouar, D. L. Mehennaoui, and P. S. Ouchtat, ”“Design, simulation and control of a
self-balancing robot in a gazebo environment and ros 2 framework ”, in Proceedings of the
International Conference on Technological Advances in Electrical Engineering (ICTAEE), Skikda, Algeria,
May 2023, pp. 98–103.
M. Boumediene, A. Maoudj, and A. L. Christensen, "HM-DRL: Enhancing Multi-Agent Pathfinding with a Heatmap-Based Heuristic for Distributed Deep Reinforcement Learning”, in Applied Intelligence, Springer, 2024.
Journal Articles
Conference Proceedings
M. Boumediene, A. L. Christensen, "Optimizing Multi-Agent Navigation with Hybrid Traffic Density Estimation”, 2024.
Digital Skills
Python
ROS/Gazebo
GIT
Docker
CAD
Wanna talk?
Contact me with any questions or just to say a few nice words.
© 2024 mouad boumediene