About

Voot Tangkaratt is a leading researcher at the intersection of reinforcement learning (RL), robotics, and statistical machine learning. His work is unified by a central challenge: enabling intelligent agents—particularly robots—to learn efficiently and robustly from limited real-world interaction. Tangkaratt made foundational contributions to sample efficiency in RL, most notably with his highly cited 2013 work on "Efficient Sample Reuse in Policy Gradients with Parameter-Based Exploration" (27 citations), which introduced a novel method for reducing variance in policy gradient estimates, a critical bottleneck for continuous control problems like robot locomotion. He has since advanced the field by tackling key problems in deep RL, including the discovery of diverse behavioral solutions through state-action mutual information maximization (21 citations), and by integrating model-based planning with dimensionality reduction (14 citations). His work also extends to practical robotic applications, such as simultaneous item picking and placing for logistics (12 citations) and efficient motor learning for humanoid robots, where he pioneered the reuse of previous experiences as simulation models. Additionally, Tangkaratt has contributed to fundamental statistical methods for conditional density estimation with feature selection (13 citations). His research consistently bridges rigorous theory with impactful, real-world robotic systems.

Research Focus

Key Achievements

6
H-Index
8
Papers
107
Total Citations
13
Avg Citations/Paper
🏆 Most Cited Paper
Efficient Sample Reuse in Policy Gradients with Parameter-Based Exploration
27 citations · 2013
📈 Most Prolific Year: 2016 (2 Papers)
🤝 Key Collaborators: 13
🏛 Institutions: Tokyo Institute of Technology, RIKEN Center for Advanced Intelligence Project, The University of Tokyo, National Institute of Information and Communications Technology

Top Papers

  1. 1
  2. 2
  3. 3
  4. 4
  5. 5
  6. 6
  7. 7
  8. 8

Key Collaborators

Contact & Links

Available for collaboration
Content generated · 14 days ago