Papers
13
Total Citations
478
H-Index
8
About
Laura Smith is a robotics and machine learning researcher whose work sits at the intersection of reinforcement learning, robot locomotion, and generalist robot control. Her research has made significant strides in making RL practical for real-world robotic systems, particularly in domains involving complex visual observations and physical environments. Her 2018 paper "SOLAR," with 131 citations, introduced deep structured representations that enabled model-based RL to operate effectively from image inputs — a foundational contribution to data-efficient robot learning. Smith has been especially influential in legged robotics, demonstrating that quadruped robots can learn robust locomotion policies directly in the real world within minutes, and that these policies can be continuously refined through fine-tuning — work that has collectively drawn over 150 citations. Her 2025 contribution to "π₀," a vision-language-action flow model for general robot control, already boasts 127 citations, signaling her growing impact on foundation models for robotics. Across her portfolio, Smith also addresses human-in-the-loop learning, meta-RL, and affordance-based manipulation, reflecting a broad commitment to building adaptable, generalizable robotic intelligence that functions reliably outside of controlled laboratory settings.
Research Focus
Key Achievements
Top Papers
- 1SOLAR: Deep Structured Representations for Model-Based Reinforcement Learning131 citations · 2018
- 2π₀: A Vision-Language-Action Flow Model for General Robot Control127 citations · 2025
- 3
- 4
- 5AVID: Learning Multi-Stage Tasks via Pixel-Level Translation of Human Videos31 citations · 2020
- 6
- 7
- 8Offline Meta-Reinforcement Learning with Online Self-Supervision8 citations · 2021
- 9
- 10