§ 01 — RESEARCH

Our work.

Technical papers from the Vertex research team — foundation models, embodied control, and the data pipelines behind extreme-purpose robotics.

RESEARCH RECORDWORKING PAPERS AND PUBLISHED PREPRINTS ARE LABELED SEPARATELY
01
Colosseum: Can a Foundation Model Make the Right Call — and Cite the Rule That Justifies It?
Romir Patel · Carnegie Mellon University
Working paper · v0.0.4 2026

A streaming-telemetry benchmark that grades doctrine-compliant choice and citation-grounded justification separately. The working paper describes 46 rules, 37 hand-authored scenarios, three executable domains, open- and closed-book studies, and an overload-mode evaluation; matching public artifact sync is pending.

02
Mesquite MoCap: Democratizing Real-Time Motion Capture with Affordable, Bodyworn IoT Sensors and WebXR SLAM
P. Vanani, D. Patel, D. Khorami, S. Munaganuru, P. Reddy, V. Reddy, B. Raghunath, I. Lallmamode, R. Patel, A. Kidané, T. Gowda
arXiv:2512.22690 2025

A low-cost, browser-deployable motion capture system that pairs bodyworn IoT sensors with WebXR-based SLAM to deliver real-time skeletal tracking without studio infrastructure — opening a path to accessible sensorimotor data collection for animation, XR interaction, and embodied-AI training pipelines.