Michael Munje

PhD student · UT Austin CS · LARG

I am a fourth-year PhD student at UT Austin, advised by Peter Stone in the Learning Agents Research Group (LARG). I study how robots and other learning agents can learn from human behavior and feedback and act effectively alongside them through continuous interactions.

Before UT, I completed an M.S. in Computer Science at Georgia Tech and received my B.S. in Computer Science from California State University, Northridge. Previously, I have interned with the ML group at NASA JPL, where I worked on detecting fresh impact craters on Mars, Microsoft (three times, optimizing advertising systems), and IBM Research (quantization for large language models). I am a regular reviewer for conferences such as NeurIPS, CVPR, AAMAS, RSS, ICRA, and CoRL.

Outside of research I enjoy playing guitar, piano, and running.

michaelmunje [at] utexas [dot] edu  ·  CV  ·  Bio  ·  Google Scholar  ·  GitHub

Michael Munje

Research

My research focuses on developing autonomous agents that can learn from and interact effectively with humans. I am particularly interested in how agents can understand human behavior, learn from the different ways people provide feedback, and adapt through interactions. Much of my current research uses social robot navigation as a domain to study these problems.

My research is guided by three main research questions:

  1. RQ1How can agents learn to understand and model human behavior? Human behavior is complex and depends on the environment and interactions with others. How can agents learn representations and predictive models of human behavior from real-world observations and use them to inform decision-making?
  2. RQ2How can agents learn from the different ways humans teach and provide feedback? Humans learn from various forms of guidance and feedback, such as demonstrations, verbal feedback, and implicit signals like facial reactions. How can agents interpret and incorporate these diverse signals into their learning process?
  3. RQ3How can agents adapt to individual humans? Human behavior and preferences can vary across individuals and environments and may change over time. How can agents learn individual preferences, adapt to these differences, and continually improve their behavior through interactions with humans?
StudentsIf you are a UT Austin student interested in my research and looking for a project to contribute to, please fill out this form.

Selected publications

STARS figure
How can agents learn to understand and model human behavior? (RQ1)
STARS: From Spatiotemporal Dynamics to Social Representations in Human-Robot Interaction

A self-supervised graph autoencoder that compresses short multi-agent interactions into compact social representations, which are strong and label-efficient for social perception and pedestrian action prediction.

Nathan Tsoi*, Michael J. Munje*, Tejas Oberoi, Rishab Maheshwari, Pengen Zheng, Tanush Chauhan, Peter Stone, Joydeep Biswas
CoRL 2026 · Acceptance rate 32.8%
SocialNav-SUB figure
How can agents learn to understand and model human behavior? (RQ1)
SocialNav-SUB: Benchmarking VLMs for Scene Understanding in Social Robot Navigation

A VQA benchmark showing that today's best VLMs still trail simple rule-based and human baselines at spatial, spatiotemporal, and social reasoning in real robot navigation scenes.

Michael J. Munje, Chen Tang, Shuijing Liu, Zichao Hu, Yifeng Zhu, Jiaxun Cui, Garrett Warnell, Joydeep Biswas, Peter Stone
CoRL 2025 · Acceptance rate 35.8%
Explainable HMT figure
How can agents adapt to individual humans? (RQ3)
Designs for Enabling Collaboration in Human-Machine Teaming via Interactive and Explainable Systems

An interactive, interpretable policy-learning platform that lets people shape their AI teammates, leading to more fluent human-AI teamwork than black-box approaches.

Rohan Paleja, Michael J. Munje, Kimberlee Chang, Reed Jensen, Matthew Gombolay
NeurIPS 2024 · Acceptance rate 25.8%

All publications

STARS: From Spatiotemporal Dynamics to Social Representations in Human-Robot Interaction

A self-supervised graph autoencoder that compresses short multi-agent interactions into compact social representations, which are strong and label-efficient for social perception and pedestrian action prediction.

Nathan Tsoi*, Michael J. Munje*, Tejas Oberoi, Rishab Maheshwari, Pengen Zheng, Tanush Chauhan, Peter Stone, Joydeep Biswas
CoRL 2026 · Acceptance rate 32.8%
SocialNav-SUB: Benchmarking VLMs for Scene Understanding in Social Robot Navigation

A VQA benchmark showing that today's best VLMs still trail simple rule-based and human baselines at spatial, spatiotemporal, and social reasoning in real robot navigation scenes.

Michael J. Munje, Chen Tang, Shuijing Liu, Zichao Hu, Yifeng Zhu, Jiaxun Cui, Garrett Warnell, Joydeep Biswas, Peter Stone
CoRL 2025 · Acceptance rate 35.8%
ComposableNav: Instruction-Following Navigation in Dynamic Environments via Composable Diffusion

A diffusion policy that learns navigation motion primitives separately and composes them at deployment to follow multi-part instructions like “overtake while staying right.”

Zichao Hu, Chen Tang, Michael J. Munje, Yifeng Zhu, Alex Liu, Shuijing Liu, Garrett Warnell, Peter Stone, Joydeep Biswas
CoRL 2025 · Acceptance rate 35.8%
Combining Open-box Simulation and Importance Sampling for Tuning Large-Scale Recommenders

Simulator-Guided Importance Sampling combines open-box simulation with importance sampling to tune large-scale ad recommenders more accurately and at lower cost.

Kaushal Paneri, Michael J. Munje, Kailash Singh Maurya, Adith Swaminathan, Yifan Shi
RecSys Workshop 2024
Designs for Enabling Collaboration in Human-Machine Teaming via Interactive and Explainable Systems

An interactive, interpretable policy-learning platform that lets people shape their AI teammates, leading to more fluent human-AI teamwork than black-box approaches.

Rohan Paleja, Michael J. Munje, Kimberlee Chang, Reed Jensen, Matthew Gombolay
NeurIPS 2024 · Acceptance rate 25.8%
Robot Air Hockey: A Manipulation Testbed for Robot Learning with Reinforcement Learning

Robot air hockey as a reinforcement learning testbed for dynamic manipulation, with multiple simulators of increasing fidelity.

Caleb Chuck, Carl Qi, Michael J. Munje, Shuozhe Li, Max Rudolph, Chang Shi, Siddhant Agarwal, Harshit Sikchi, Abhinav Peri, Sarthak Dayal, Evan Kuo, Kavan Mehta, Anthony Wang, Peter Stone, Amy Zhang, Scott Niekum
ICRA Workshop 2024
Using Machine Learning to Reduce Observational Biases When Detecting New Impacts on Mars

A crater-detection model deployed across nearly the entire surface of Mars that reduces observational bias and found many previously unknown new impacts.

Kiri L. Wagstaff, Ingrid J. Daubar, Gary Doran, Michael J. Munje, Valentin T. Bickel, Annabelle Gao, Joe Pate, Daniel Wexler
Icarus 2022
TEAM3 Challenge: Tasks for Multi-Human and Multi-Robot Collaboration with Voice and Gestures

A challenge suite of tasks for collaboration among multiple humans and multiple robots using voice and gestures.

Michael J. Munje, Lylybell K. Teran, Bradon Thymes, Joseph P. Salisbury
HRI LBR 2023
Providing Predictions of Adversary Movements in a Gridworld Environment to a Human-Machine Team Improves Teaming Performance

Giving a human-machine team theory-of-mind predictions of an adversary's movements improves their teaming performance.

Jeffry A. Coady, Paul Dysart, Aidan Schumann, Stephan A. Koehler, Michael J. Munje, William D. Casebeer, David M. Huberdeau
SPIE DCS 2023