Bhavya Sukhija
Papers
2
Total Citations
17
H-Index
2
About
Bhavya Sukhija is a rising star in robotics and machine learning, whose research tackles the fundamental challenge of enabling autonomous systems to learn safely and efficiently in the real world. Her work sits at the intersection of safe reinforcement learning, model-based control, and meta-learning, with a focus on developing algorithms that can adapt to new environments without costly failures. Sukhija’s most cited paper, "GoSafeOpt" (2023, 11 citations), introduces a scalable approach to safe exploration for global optimization of dynamical systems—a critical advance for applying learning directly on physical hardware, where a single mistake can cause damage. She further demonstrates her impact with "PACOH-RL" (2024, 6 citations), a probabilistic model-based meta-reinforcement learning algorithm that enables rapid, data-efficient adaptation to changing dynamics. This work is particularly notable for bridging the gap between sample efficiency and safety, a key bottleneck in deploying RL in robotics. With her focus on principled, uncertainty-aware methods, Sukhija is shaping a future where robots can learn robustly from limited interaction, making her a researcher to watch for students and practitioners interested in the next generation of intelligent, real-world systems.
Research Focus
Key Achievements
Top Papers
- 1
- 2