| This project is concerned with the in-field autonomous operation of unmanned marine vehicles in accordance with convention for safe and proper collision avoidance as prescribed by the Coast Guard Collision Regulations (COLREGS). These rules are written to train and guide safe human operation of marine vehicles and are heavily dependent on human common sense in determining rule applicability as well as rule execution, especially when multiple rules apply simultaneously. To capture the flexibility exploited by humans, this work applies a novel method of multi-objective optimization, interval programming, in a behavior-based control framework for representing the navigation rules, as well as task behaviors, in a way that achieves simultaneous optimal satisfaction. We validate this approach using multiple autonomous surface craft. |
| Interval programming (IvP) is a model and set of algorithms for representing and solving multi-objective optimization problems. A common thread in the applications motivating the development of IvP is the need to fully automate a decision (as in a robot or autonomous vehicle), or provide a decision to a human in a critical window of time with little or no room for human interaction (as with decision aids in a submarine combat control center). Furthermore, the environment is typically composed of distinct components, or ``sub-situations'' that, in isolation, are fairly familiar and easy to handle. Their combination however typically creates a completely unique and unfamiliar overall situation where the decision-making challenge is in finding the right balance between the concerns of each sub-situation. |
| In this work, multi-objective decision making techniques are utilized to address the problem of controlling an autonomous physical agent in real-world environments that present simultaneous competing navigation and sensing objectives. The long-term objective of this work is to dramatically advance the capability of an autonomous agent to act effectively in a physical environment. In short, the agent must make good decisions with good information. In practice, the quality of the information is tied not only to the quality of the sensors and sensor algorithms, but also to the manner in which the agent positions and utilizes those sensors. Doing the latter effectively, is often in contradiction or competition with agent actions that more directly relate to mission objectives. The common thread addressed in this work is the ability to make good control decisions that balance the multiple needs of sensing and acting in real environments. |