Kartik Nagpal

Kartik Nagpal

Learning complex robot behavior for safety-critical environments

Sitemap

A list of all the posts and pages found on the site. For you robots out there is an XML version available for digesting as well.

Pages

Posts

Leveraging

less than 1 minute read

Published:

I have taken so many classes in my academic career, but I didn’t really understand how much work goes into teaching a class till I helped TA this graduate class. My advisor, Negar Mehr, was teaching a special topics class in the UC Berkeley mechanical engineering department, and asked if I would be interested in being a Graduate Student Instructor (GSI) for the course. As I hope to eventually be a professor, I really wanted this chance to peel away the curtain and experience

Lead TA for ME290 Modeling and Control of Multi-Agent Systems

less than 1 minute read

Published:

I have taken so many classes in my academic career, but I didn’t really understand how much work goes into teaching a class till I helped TA this graduate class. My advisor, Negar Mehr, was teaching a special topics class in the UC Berkeley mechanical engineering department, and asked if I would be interested in being a Graduate Student Instructor (GSI) for the course. As I hope to eventually be a professor, I really wanted this chance to peel away the curtain and experience what it takes to run a class.

news

Stillwell_fellowship

Published:

I was awarded the Stillwell Fellowship by the Department of Aeronautical Engineering at UIUC! I was also honored with an supplementary Robert Beatty Fellowship!

Goddard_offer

Published:

I was offered a summer internship at NASA Goddard Space Flight Center with a focus on “Decision Making for Collision Avoidance via Machine Learning”!

Nsf_grfp

Published:

I was awarded the National Science Foundation Graduate Research Fellowship (NSF GRFP)!

Nasa_ai_showcase

Published:

I was invited to talk about my work at the NASA AI Showcase at Goddard Space Flight Center! This included an hour-long oral presentation to a crowd of over 50 NASA engineers and scientists, followed by a poster session.

Gsi_class

Published:

For the Fall 2024 semester, I got the opportunity to act as the teaching assistant for my advisors’ “ME 292B/EE290: Modeling and Control of Multi-Agent Systems” special topics course.

Goddard_completed

Published:

I completed my NASA intership! I developed a novel transformer architecture for space weather prediction, along with its effects on satelite orbit perturbation. This was then trained on real satellite data to improve uncertainty quantification efforts.

Prelim_exam

Published:

I passed my UC Berkeley Ph.D Preliminary Exam!

Ncas

Published:

I was invited for an oral presentation and poster session at the Northern California Aerospace Symposium (NCAS) 2025!

Bair_robotics_lightning_talk

Published:

I delivered a lightning talk at the Spring 2025 BAIR Robotics Workshop: The Dawn of a General Purpose Robotics Era!

portfolio

publications

Transparency and Accountability in Space Domain Awareness: Demonstrating ASTRIAGraph’s Capabilities with the United Nations Registry Data

Published in Texas ScholarWorks, 2021

Near-earth space is geopolitically and commercially contested, and in need of environmental protection. To achieve space safety, security, and sustainability, we are developing ASTRIAGraph, a framework that enables monitoring, assessment, and verification of space actor behavior in the context of legal and policy instruments.

Download here

Optimal Robotic Assembly Sequence Planning (ORASP): A Sequential Decision-Making Approach

Published in 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2024



Proposed a new formulation of the robotic assembly sequencing problem as a Markov Decision Process. Then showed how a class of methods called Graph Exploration Assembly Planners (GEAPs) can be used to gather optimal assembly sequences from a graph. We then showcased a deep reinforcement learning extension for handling very complex structures, all while handling diverse constraints.

Download here

Leveraging Large Language Models for Effective and Explainable Multi-Agent Credit Assignment

Published in 2025 International Conference on Autonomous Agents and Multiagent Systems (AAMAS), 2025



Robots collaborating for shared goals face a challenge: determining each agent’s contribution to team success or failure. This paper introduces LLM-MCA and LLM-TACA, methods that use Large Language Models to assign credit and even tasks to individual agents, significantly outperforming existing approaches.

Download here

talks

teaching