Alper Kamil Bozkurt
Papers
2
Total Citations
19
H-Index
2
About
Alper Kamil Bozkurt is a leading researcher at the intersection of control theory, reinforcement learning, and robotics security. His work focuses on developing resilient autonomous systems capable of operating safely in stochastic and adversarial environments. Bozkurt’s most impactful contribution addresses a critical vulnerability in modern robotics: the problem of secure planning against stealthy attacks on control signals. In his highly cited 2020 paper (17 citations), he pioneered a model-free reinforcement learning framework that enables robots to plan secure trajectories even when actuators are compromised by an attacker with full knowledge of the controller and intrusion-detection system—a significant advance for safety-critical applications. More recently, in 2024, Bozkurt has tackled the limitations of Decision Transformers in stochastic robotics settings, introducing a novel method that steers these powerful offline RL models via temporal difference learning to improve their robustness. His work bridges theoretical rigor with practical deployment challenges, earning recognition for advancing the security and reliability of learning-based control systems. Bozkurt’s research continues to shape how autonomous agents make decisions under uncertainty and threat.
Research Focus
Key Achievements
Top Papers
- 1
- 2Steering Decision Transformers via Temporal Difference Learning2 citations · 2024