UTSeaSim - UT Austin's Naval Surface Navigation Simulator

UTSeaSim is a custom-designed naval surface navigation simulator that was designed to realistically model ship movement constraints in marine environments. UTSeaSim is written as a supporting platform for research on autonomous sea navigation. It uses realistic 2D physical models of marine environments and sea vessels, and runs both in GUI and in non-GUI modes.

The simulator's core contains three main modules:

Sea Environment module
Ship module
Decision-Making module

The sea environment module includes models of winds, water currents, waves, and obstacles. The ship module models all relevant aspects of a ship, including the ship's physical properties, sensing capabilities, and ship actuators. The decision making-module implements an agent that controls a ship autonomously. At each time step, the agent receives the perceptions sensed by the ship, processes them to update its current world state, and decides on control actions for the ship based on its current world state and its decision-making strategy.

All of the above components are plug-and-play: each one can easily be replaced by a component of the same type that uses alternative implementation. This includes the agent's perception processing algorithms, world model, and decision making strategy; the ship's model; and the different environmental models.

UTSeaSim supports communication between ship-controlling agents, as well as communication between agents and the simulation environment (for instance, for sending and collecting statistics).

Papers and Videos

UTSeaSim - Example Patrol Tasks.
Download video: (mp4)

We used UTSeaSim when solving a problem of frequency based patrol in marine environments. We proved the problem to be NP-Hard, and presented a heuristic algorithm that performs well in practice. Full details of our approach are available in the following article.

Multiagent Patrol Generalized to Complex Environmental Conditions
Noa Agmon, Daniel Urieli, and Peter Stone.
AAAI, 2011.
Click here for the paper's web page including pdf and ps versions, and bibtex.

Complementing paper materials:
- A poster that was presented in AAMAS '11: pdf, pptx.
- A few videos demonstrating our algorithm on a simple case.
  - Ships circling in a patrol mission when there are no currents: mp4 avi ogv
  - Adding currents makes this patrol non-optimal with respect to minimizing the frequency of visits in a points: mp4 avi ogv
  - Applying our algorithm, a better patrol strategy is found: mp4 avi ogv