Abhishek Paudel
CS PhD Candidate, George Mason University
I am a PhD candidate in Computer Science at George Mason University, advised by Prof. Gregory J. Stein in the Robotic Anticipatory Intelligence & Learning (RAIL) Group.
My research is at the intersection of robotics and machine learning, with a focus on long-horizon planning under uncertainty. My work focuses on integrating planning and learning via model-based approaches to enable effective and reliable long-horizon decision-making under uncertainty for problems such as navigation, multi-robot coordination, task planning, deployment-time introspection and adaptation, and planning with foundation models.
I graduated with a degree in computer engineering from Tribhuvan University, IOE, Pulchowk Campus in Nepal. Before enrolling at GMU, I worked with UNICEF Nepal on several SMS-based platforms including U-Report Nepal. I was also a Microsoft Student Partner and a Fusemachines AI Fellow.
news
| Jan 31, 2026 | Our paper Multi-Robot Learning-Informed Task Planning Under Uncertainty was accepted at ICRA 2026. |
|---|---|
| Dec 2, 2025 | Successfully defended my PhD dissertation titled Robots that Introspect: Improving Deployment-Time Performance for Long-Horizon Planning under Uncertainty. |
| Nov 14, 2025 | I attended the Virginia Robotics Symposium, where I presented a poster titled Learning and Introspection for Long-Horizon Robot Planning uncer Uncertainty. |
| Oct 7, 2025 | I gave an invited talk in the Computing@PNNL Seminar at Pacific Northwest National Laboratory. |
| Apr 18, 2025 | Our paper Deployment-time Selection of Prompts for LLM-informed Object Search in Partially-Known Environments was accepted at ICRA FMNS Workshop 2025. [Paper] |
| Apr 17, 2025 | I was selected to ICRA 2025 Doctoral Consortium. I will present my work on Introspection for Long-Horizon Robot Planning under Uncertainty. [Abstract] |
research overview
Task Planning under Uncertainty
Developing planning abstractions that allow robots to efficiently perform complex tasks despite uncertainty and incomplete knowledge about their environment, e.g. missing objects and unknown locations, while performing other task-relevant actions in large, uncertain environments.
Multi-Robot Coordination
Coordinating teams of robots to efficiently divide search efforts and complete complex tasks while acting concurrently in large, uncertain environments. Planning and coordination for heterogenous teams (drones, rovers, humanoids) whose skills, efficiency, and capabilities may vary.
Introspection for Long-Horizon Planning
Developing introspection for robots—the ability to look back at past decisions and imagine how alternative behaviors could have led to better outcomes via counterfactual reasoning without first-hand experience of failure.
Autonomous Navigation and Adaptation
Enabling robots to efficiently navigate unknown environments via integrated planning and learning methods. Developing techniques for reliable deployment-time adaptation, allowing robots to monitor their own performance and select the best adaptation strategies during deployment for reliable navigation in novel environments.
Foundation Models for Planning
Combining the commonsense world knowledge of LLMs, VLMs and emerging foundation models with classical model-based planning for effective and reliable long-horizon decision-making.