Search and Rescue is a task where the rescuers need to be cognitively agile, strategically consistent, and efficient to save as many trapped victims as possible. In a team scenario, the rescuers must additionally coordinate with each other based on their individual skillsets. However, in such a dynamic scenario, the rescuers’ mental models may get outdated soon, and it may be difficult to coordinate with other rescuers under a time constraint and cognitive overload. This is where AI assistance can be helpful to update the rescuers’ mental models.
In this thesis, I discuss how to develop agents based on Machine Theory of Mind (MToM) to infer the beliefs, intentions, and desires of the rescuers from their observations and actions. By generating a mental model, an AI agent can intervene when it detects a rescuer might act based on a false belief. We create neural agents that predict navigation and rescue strategies from ongoing trajectories of rescuers and compare them with human predictors. We also study transfer learning approaches on using data from a smaller map to learn robust neural agents. Finally, we study the paradigm of imitation learning to learn policies from expert trajectories and propose a sample efficient method over existing baselines.
Katia Sycara (Advisor)
Zoom Participation. See announcement.