Cong-Tinh Dao
Papers
1
Total Citations
3
H-Index
1
About
Cong-Tinh Dao is a researcher advancing the frontier of intelligent robotic control through deep reinforcement learning (DRL). His work addresses a critical bottleneck in deploying DRL in real-world settings: the need for smooth, physically feasible actions. In his highly cited 2024 paper, "Gradient-based Regularization for Action Smoothness in Robotic Control with Reinforcement Learning," Dao introduces a novel regularization technique that penalizes abrupt changes in action sequences. This contribution directly mitigates the jerky, unstable behaviors often seen in learned policies, making them safer and more efficient for physical robots. By integrating this gradient-based penalty into standard DRL algorithms, his method provides a principled way to enforce action smoothness without sacrificing task performance. Though early in its citation lifecycle, this work is already recognized as a key step toward bridging the gap between simulated DRL successes and reliable real-world deployment. Dao’s research sits at the intersection of control theory, optimization, and machine learning, offering practical solutions for autonomous systems ranging from robotic arms to mobile platforms.
Research Focus
Key Achievements
Top Papers
- 1