Auburn statistician receives award from the Office of Naval Research to develop algorithms to help optimize naval navigation
Elvan Ceyhan, associate professor in the Department of Mathematics and Statistics, is the recipient of an award sponsored by the Department of the Navy, Office of Naval Research (ONR) under award number N00014-22-1-2572 for the project Adversarial Risk Analysis for Optimal Obstacle Evasion. The project, a collaboration with David Banks of Duke University, received an award of $358,000 with Auburn receiving $214,000 and Duke receiving $144,000.
The overall goal of the research project is to create procedures that avoid adversarial threats for continuous paths such as ships navigating the ocean. It will result in algorithms and methodology in Adversarial Risk Analysis (ARA).
“Imagine a grid over the ocean with traversable edges and diagonals,” said Ceyhan. “This network grid superimposed on the ocean’s surface helps us to unlock approximate paths for naval ships to safely navigate.”
“Adversarial Risk Analysis looks at a navigating agent and an adversarial agent, which would be capable of putting obstacles in the path,” said Ceyhan. “With this work, we can minimize the potential damage and expense on the path of the navigating agent to reach its required destination.”
The proposed approach is a combination of two concepts: Canadian traveler’s problem (CTP) in optimization and ARA. CTP was motivated by the traversal strategies in harsh winter conditions in Canada. “Think of a traveler who needs to reach a target, say, a pharmacy or a supermarket from her home in a vehicle,” Ceyhan explained. “However, some streets may be blocked (due to heavy snow), and she needs to navigate through the street which forms the traversal grid.”
The current project, which proposes a relatively new concept for finding feasible solutions to problems through both basic and applied research, is an adaptation of CTP to weighted and partially blocked spatial networks with adversarial agents on the network.
“We will first define the problem specification and complete the Adversarial Risk Analysis,” he said. “In the next phase of the project, we will use reinforcement learning to find the policy that is nearly optimal.”
In addition to life-saving research in naval navigation, this work also has real-world applications in both robotics and inventory allocation.
“Operations research is filled with finding solutions to traversal problems hindered with obstacles,” he added. “This work can help operations management find more successful solutions to overcome such challenges.”
BY MARIA GEBHARDT
Any opinions, findings, and conclusions or recommendations expressed in this material are those of the author(s) and do not necessarily reflect the views of the Office of Naval Research.