Biography

Hello! I am a Ph.D. candidate at the Robotics Institute at Carnegie Mellon University. I am advised by Prof. Oliver Kroemer and work in close collaboration with Prof. Andrea Bajcsy.

My research lies at the intersection of robot learning and control, with a focus on integrating semantic, structural, and algorithmic priors to develop generalizable and safe robot policies. I am particularly interested in leveraging large-scale foundation models for semantic scene understanding to perform embodied tasks in novel environments as well for enabling robots to understand and adhere to semantic safety constraints in dynamic and interactive settings. My long-term goal is to develop robots that can operate safely in unstructured environments, perform complex dynamic and interactive tasks, and collaborate with humans in assistive and caregiving roles.

Prior to joining the Ph.D. program, I worked under the supervision of Prof. Howie Choset and Dr. Matthew Travers on developing a belief space planning framework for utilizing contacts to reduce uncertainty.

---

Research

Safe Manipulation via Task-Relevant Reach-Avoid Reinforcement Learning (Under review)

Saumya Saxena, Andrea Bajcsy, and Oliver Kroemer. 2025 [Paper]

---

GraphEQA: Using 3D Semantic Scene Graphs for Real-time Embodied Question Answering (Under review)

Saumya Saxena, Blake Buchanan, Chris Paxton, Bingqing Chen, Narunas Vaskevicius, Luigi Palmieri, Jonathan Francis, and Oliver Kroemer. arXiv, 2024 [arXiv] [Website]

---

MResT: Multi-Resolution Sensing for Real-Time Control with Vision-Language Models

Saumya Saxena, Mohit Sharma, and Oliver Kroemer. Conference on Robot Learning (CoRL), 2023 [arXiv] [Website]

---

Dynamic Inference on Graphs using Structured Transition Models

Saumya Saxena and Oliver Kroemer. International Conference on Intelligent Robots and Systems (IROS), Oct 2022. [arXiv] [Video]

---

Search-Based Task Planning with Learned Skill Effect Models for Lifelong Robotic Manipulation

Jacky Liang, Mohit Sharma, Alex LaGrassa, Shivam Vats, Saumya Saxena and Oliver Kroemer. International Conference on Robotics and Automation (ICRA), May 2022. [Paper] [Website]

---

Learning Reactive and Predictive Differentiable Controllers for Switching Linear Dynamical Models

Saumya Saxena, Alex LaGrassa and Oliver Kroemer. International Conference on Robotics and Automation (ICRA), May 2021. [arXiv] [Video]

---

Learning Active Task-Oriented Exploration Policies for Bridging the Sim-to-Real Gap

Jacky Liang, Saumya Saxena, Oliver Kroemer. Robotics: Science and Systems (RSS), July 2020. [arXiv] [Website]

---

A Real-time Augmented Reality Surgical System for Overlaying Stiffness Information

Nicolas Zevallos, Arun Srivatsan Rangaprasad, Hadi Salman, Lu Li, Jianing Qian, Saumya Saxena, Mengyun Xu, Kartik Patath and Howie Choset. Proceedings of Robotics: Science and Systems (RSS), June 2018. [Paper]

---

A surgical system for automatic registration, stiffness mapping and dynamic image overlay

Nicolas Zevallos, Arun Srivatsan Rangaprasad, Hadi Salman, Lu Li, Jianing Qian, Saumya Saxena, Mengyun Xu, Kartik Patath and Howie Choset. Proceedings of International Symposium on Medical Robotics (ISMR), March 2018. [arXiv] [Video]

---

Technical Reports

Motion planning under uncertainty and sensing limitations using exploration versus exploitation

Saumya Saxena, Matthew Travers, Howie Choset. Masters Thesis report, August 2019. [Report]

---

Coverage Using Graph-based Motion Planning for Autonomous Exploration Of Non-uniform Environments

Saumya Saxena. May 2018. [Report] [Video1] [Video2]

---

CatAstroφ - Inertial reorientation of a freely falling cat using non-holonomic motion planning

Alexander Duncan, Andreas Fatschel, Dimitrios Nikitopoulos, Saumya Saxena, Steven Crews. May 2018. [Report] [Video]