Autonomous decision-making
POMDPs, Monte Carlo tree search, trajectory prediction, and interactive planning for complex traffic.
Welcome.
I am a Lecturer in the Department of Mechanical Engineering at İstanbul University–Cerrahpaşa. My research combines reinforcement learning, model-based control, and interpretable AI to make autonomous vehicles and robots safer, more capable, and easier to validate.
I completed my PhD at the University of Surrey in 2025, where I worked on intelligent decision-making for autonomous driving. My background spans academic research and automotive engineering, including roles with Innovate UK projects, Sellafield Game Changer, Ford Otosan, and AVL.
Research
My work sits at the intersection of control engineering, artificial intelligence, robotics, and intelligent mobility.
POMDPs, Monte Carlo tree search, trajectory prediction, and interactive planning for complex traffic.
Safe policy learning, behavioural cloning, human-like driving, and multi-agent coordination.
Neuro-symbolic learning and language-model reasoning for explainable, knowledge-informed policies.
Model predictive control, feedback control, and hybrid model-based / learning-based architectures.
Closed-loop testing, scenario evaluation, digital twins, and reproducible benchmarking.
Minimal-sensor reinforcement learning and flight-dynamics simulation for autonomous thermal soaring.
Currently
Working on safe and interpretable driving policies, minimal-sensor autonomous soaring, and multi-agent learning for coordinated vehicle and robot control.
Selected projects
Research theme
Learning, planning, and neuro-symbolic reasoning for interactive traffic, with safety and interpretability at the centre of evaluation. The original top-down visualisation is inspired by HighD-style trajectory analysis; it is not recorded dataset footage.
Related publicationOpen model · 2026
An open MATLAB/Simulink model for traction, braking, resistive forces, and speed-control studies.
RL · Flight dynamics
Minimal-sensor reinforcement learning for energy-efficient soaring in JSBSim, with FlightGear visualisation.
Digital twin · Robotics
Warehouse digital-twin and load-balancing research developed through industry-facing Innovate UK work.
Selected publications
I. Sharifi, M. Yıldırım, and S. Fallah
Transportation Research Record, 2680(1), 416–431
M. Yıldırım, B. Dagda, V. Asodia, and S. Fallah
Robotics and Autonomous Systems, 193, 105114
I. Sharifi, M. Yıldırım, and S. Fallah
Applied Sciences, 15(23), 12464
A. Özbeyaz, M. Yıldırım, and F. Tufaner
Discover Applied Sciences, 7, 19
M. Yıldırım, S. Mozaffari, L. McCutcheon, M. Dianati, A. Tamaddoni-Nezhad, and S. Fallah
IEEE 25th International Conference on Intelligent Transportation Systems (ITSC)
M. Yıldırım
SAE Technical Paper 2017-01-1064
Behavioral Cloning Models Reality Check for Autonomous Driving · arXiv, 2024
Human-Like Autonomous Driving on Dense Traffic · arXiv, 2023
Otonom Araçlarda Kontrol Yöntemleri · Book chapter, 2025
Experience & education
Teaching, supervision, and research in autonomous systems, control, reinforcement learning, and engineering simulation.
AI-enabled mobile-robot autonomy, warehouse digital twins, load balancing, and navigation concepts for safety-critical environments.
Intelligent decision-making methods for autonomous driving using reinforcement learning, prediction, imitation learning, and online planning.
Product development, powertrain NVH simulation, engine-load modelling, modal analysis, and structural verification.
University of Surrey
İstanbul Technical University
Sakarya University
Teaching & supervision
I supervise undergraduate theses and design projects in autonomous driving, reinforcement learning, simulation, control, and traffic-data analysis.
Discuss a projectContact
I welcome conversations about international research collaboration, student projects, and postdoctoral or visiting-researcher opportunities.
mustafa.yildirim@iuc.edu.tr