Hemant Kumawat
Scholar

Hemant Kumawat

Google Scholar ID: 2iUnwBwAAAAJ
Georgia Institute of Technology
Deep LearningRoboticsReinforcement LearningComputer Vision
Citations & Impact
All-time
Citations
99
 
H-index
5
 
i10-index
2
 
Publications
12
 
Co-authors
8
list available
Resume (English only)
Academic Achievements
  • Paper 'AdaCred: Adaptive Causal Decision Transformers with Feature Crediting' accepted to AAMAS 2025 (Dec 2024).
  • Paper 'RoboKoop: Efficient Control Conditioned Representations from Visual Input in Robotics using Koopman Operator' accepted to CORL 2024 (Sep 2024).
  • Paper 'STEMFold: Stochastic Temporal Manifold for Multi-Agent Interactions in the Presence of Hidden Agents' accepted to L4DC 2024 (Jul 2024).
  • Published research in top-tier venues including CORL, L4DC, AAMAS, IMS, and IJCNN.
  • Serving as reviewer for NeurIPS 2024, ICLR 2025, ICML 2025, AISTATS 2024, and IJCNN 2024.
Background
  • Robotics researcher focused on building intelligent, generalizable autonomous agents that operate in complex physical environments.
  • Explores the intersection of Robotics, Generative Modeling, Computer Vision, and Reinforcement Learning, with emphasis on learning from limited supervision and unstructured data.
  • PhD work bridges perception, prediction, and control through compositional and causal ML models for robust decision-making in real-world scenarios.
  • Has a proven track record of building production-scale ML systems integrating perception, prediction, and control for autonomous agents.
  • Previously led a 20-member student autonomous vehicle team, deploying full perception, planning, and control stacks on real hardware.