Research Publications

Explore our latest research findings, scientific studies, and white papers contributing to ocean science.

Autonomy, Simulation 2025

Toward standard interfaces for high-level Autonomy Simulation with MOOS-IvP

Mathew Charles Schwartzman

Published in ICRA 2025 AQ2UASIM Workshop

Abstract

Digital simulation of robotic systems is a near-universal need in the development of new autonomy. That need is exponentially increased in the case of oceanographic engineering, which remains one of the most expensive applications of the technology. While many modern robotics simulators aim to implement high-fidelity physics engines and 3D graphics, this paper highlights the conversely lightweight toolset used in the MOOS-IvP ecosystem for high-level and multi-vehicle simulation. Additionally we evaluate a modular container stack, WebMOOS, containing an API and other microservices that serve a set of standard interfaces for new engineers and scientists.

Download PaperDownload Poster
Autonomy, Swarm Robotics 2025

A modular interface for multi-agent Marine Autonomy Simulation

Mathew Charles Schwartzman, Charlie Benjamin

Published in Oceans 2025 Great Lakes

Abstract

Marine autonomy is a rapidly-growing field, with wide-ranging applications in climate health monitoring, ocean exploration, and naval security. As the technologies mature, scaling is typically limited by the rate at which engineers can evaluate configuration parameters and autonomy logic in the field and in simulation. With field time already a scarce commodity, there exists a need for rapidly-repeatable tests in a reliable and flexible digital ocean twin. Such simulated tests need to be executable with a reasonable balance of fidelity, mutability, and speed. This paper presents a simulation framework that focuses on meeting these criteria within a modular, robust, and accessible software package using MIT’s MOOS-IvP middleware. MOOS is a lightweight middleware designed to support interprocess communication in large, multi-application robotic platforms like AUVs (Autonomous Underwater Vehicles) and ASVs (Autonomous Surface Vehicles). Its extension, MOOS-IvP (Interval Programming), was developed around the paradigm of behavior-based autonomy, using high-level objective functions to rapidly and adaptively control speed, heading, and depth. MOOS-IvP also provides a very simple simulator and User-Interface to test these behaviors in faster-than-realtime environments. This empowers users to rapidly spin up test missions with entire fleets of autonomous agents for the purposes of fine-tuning configurations and vehicle dynamics. This paper presents a framework of thin tools along this paradigm which provides a more standard interface to the MOOS-IvP simulator and middleware. This framework, WebMOOS, aims to use modern industry standards in microservice and web architecture, along with common devops techniques like containerization to streamline the sim-to-real pipeline with science users in mind.