Home /Research /CataBotSim: A Realistic Aquatic Simulator for Autonomous Surface Vehicle Testing
PERCEPTION

CataBotSim: A Realistic Aquatic Simulator for Autonomous Surface Vehicle Testing

Dinithi Silva-Sassaman, Mingi Jeong, Paul Sassaman, Zongcai Wu, Alberto Quattrini Li

Year
2024
Citations
3

Abstract

While there are many self-driving car simulators, there is a general lack of high-fidelity aquatic simulators that allow testing perception and navigation pipelines for autonomous surface vehicles (ASVs). We present an open source, user friendly, and realistic Unity3D-based aquatic simulator for ASVs, called CataBotSim, which includes support for multiple ASV models, on-board sensors, such as LiDAR, Marine Radar, IMU, GPS, and cameras, as well as a number of different objects that can act as static or dynamic obstacles. To achieve high-fidelity simulations, we incorporated environmental effects like rain, currents, and wind, which impact sensor measurements and/or ASV motions as observed in real-world environments. We included interfacing with the Robot Operating System (ROS). We also provided tools to generate real-world scenes, create datasets, record navigation metrics, and provide ground truth perception data to ease robotic simulation experiments. The project code for the CataBotSim can be found at https://github.com/dartmouthrobotics/catabot_unity/. A demonstration video of CataBotSim is available at https://youtu.be/Rgfkm352ho4.

Keywords

SimulationComputer scienceMarine engineeringEngineering

Related papers

Browse all PERCEPTION papers