Amir Mazaheri
Papers
2
Total Citations
28
H-Index
2
About
Amir Mazaheri is a researcher advancing the robustness of deep visuomotor policies for robotic manipulation. His primary focus lies at the intersection of computer vision, reinforcement learning, and robot control, specifically addressing the fragility of learned policies under physical disturbances. His most influential work, "Pay Attention! - Robustifying a Deep Visuomotor Policy Through Task-Focused Visual Attention" (2019, 26 citations), introduces a novel attention mechanism that enables a robot manipulator to maintain stable object grasping even when subjected to accidental or adversarial bumps. By forcing the policy to focus on task-relevant visual features—rather than the entire scene—Mazaheri’s approach significantly improves policy resilience without requiring additional training data or complex sensor suites. This contribution is particularly notable for its practical impact: it offers a lightweight, modular solution that can be integrated into existing deep visuomotor architectures. Through this work, Mazaheri demonstrates how targeted visual attention can bridge the gap between simulation-trained policies and real-world deployment, making robotic systems more reliable in dynamic, unpredictable environments. His research continues to inspire new directions in robust, attention-driven robot learning.
Research Focus
Key Achievements
Top Papers
- 1
- 2