Hey, I'm Max!

MS in Computer Science at Cornell University

Max Pace

I'm an AI Engineer at Scout AI building robotic foundation models for defense. I got my Master of Science degree studying robot learning at Cornell University under professors Sanjiban Choudhury and Wei-Chiu Ma. My work spans training VLAs for offroad driving and learning robotic manipulation from videos of humans. I'm also very excited about entrepreneurship and am always looking for new ideas to build and grow. Please feel free to reach out if you would like to talk with me about anything!

Autonomous offroad driving

Cornell Mars Rover

Learning from Human Videos

Publications

X-Diffusion: Training Diffusion Policies on Cross-Embodiment Human Demonstrations
Maximus Adrian Pace, Prithwish Dan, Chuanruo Ning, Atiksh Bhardwaj, Moyun Du, Edward Duan, Wei-Chiu Ma, Kushal Kedia
International Conference on Robotics and Automation (ICRA), 2026. [Website]
X-Diffusion co-trains diffusion policies on cross-embodiment using a classifier to maximally leverage human data without learning infeasible behaviors.
X-Sim: Cross-Embodiment Learning via Real-to-Sim-to-Real
Prithwish Dan, Kushal Kedia, Angela Chao, Edward Duan, Maximus Adrian Pace, Wei-Chiu Ma, Sanjiban Choudhury
Oral Presentation at 9th Annual Conference on Robot Learning (CoRL), 2025.[Website]
X-Sim learns robot actions from action-less human videos using an object-centric reward in simulation, bridging the embodiment gap while increasing data efficiency and generating diverse synthetic data.
Rhyme
One-shot imitation under mismatched execution
Prithwish Dan, Kushal Kedia, Angela Chao, Maximus Adrian Pace, and Sanjiban Choudhury
International Conference on Robotics and Automation (ICRA), 2025. [Website]
RHyME introduces a new framework that enables robots to learn from watching human demonstrations even when there are differences in execution styles.
APRICOT
APRICOT: Active Preference Learning and Constraint-Aware Task Planning with LLMs
Huaxiaoyue Wang, Nathaniel Chin, Gonzalo Gonzalez-Pumariega, Xiangwan Sun, Neha Sunkara, Maximus Adrian Pace, Jeannette Bohg, Sanjiban Choudhury
8th Annual Conference on Robot Learning (CoRL), 2024. [Website]
APRICOT combines the generative ability of LLMs with Bayesian active preference learning, allowing robots to interactively query users to reduce uncertainty.
MOSAIC
MOSAIC: A Modular System for Assistive and Interactive Cooking
Huaxiaoyue Wang, Kushal Kedia, Juntao Ren, Rahma Abdullah, Atiksh Bhardwaj, Angela Chao, Kelly Y Chen, Nathaniel Chin, Prithwish Dan, Xinyi Fan, Gonzalo Gonzalez-Pumariega, Aditya Kompella, Maximus Adrian Pace, Yash Sharma, Xiangwan Sun, Neha Sunkara, Sanjiban Choudhury
8th Annual Conference on Robot Learning (CoRL), 2024. [Website]
MOSAIC combines large pre-trained models for general tasks with task-specific modules to enable collaborative cooking.
Video2Demo
Video2Demo: Grounding Videos in State-Action Demonstrations
Yash Sharma, Yuki Wang, Kelly Chen, Maximus Pace, Sanjiban Choudhury
Video2Demo combines GPT-4 with vision-language models to convert human video demonstrations into temporally coherent state-action sequences for robot imitation learning.