Estimating and Calibrating Situation Awareness for Improving Human-Robot Teaming Performance

Project begins 2022.

The objective of this research project is to develop methods for estimating and calibrating situational awareness in teams to optimize their overall performance. We hypothesize that there is a desired level of situation awareness (SA) for each human agent on the team that will create the team SA or “common picture”, and that this desired level depends on the context or environment in which the team is operating. We treat SA as a dynamic variable that can improve as more information becomes available and degrade over time, or decrease quickly as the environment changes.

Since the ultimate goal is to improve the team performance, we must first define the task and criteria for its successful accomplishment. As a starting point, we will consider a search and reconnaissance task in an unknown outdoor environment with one or two humans and up to five UGVs. As the research develops, we will increase the complexity of the task to add logistics or convoy tasks, and also add UAVs to the autonomous vehicles. The goal will be to explore and map the entire area, finding any unusual features (e.g., discards from previous occupants) while avoiding any dangers (e.g., ditches, cliffs, land mines), and complete the task in the minimum amount of time. The quality of the map, related to the density of the search, may also be considered as a performance criteria.

The research questions that we pose are:

RQ1: What information about the agents and their contexts has to be known and communicated to (which) other agents to support team SA?

RQ2: How much information is needed to improve team SA and team performance, and under what conditions does more information degrade team SA?

RQ3: How does the SA of the team members (both humans and robots) relate to the overall performance of the team? Is there a desired level of SA?

Information sharing is known to improve SA, but information overload can decrease SA. Prior work showed how SA (of the human) can increase trust (in the robot) and overall team performance. In larger human-robot teams, each team member should have an appropriate level of SA regarding the other team members, in order to accomplish the team’s goals effectively. SA of robots is not well-defined, SA of human teams is still poorly understood, and thus defining SA in human-agent teams, and how it relates to overall performance, is an open question and requires basic research.

By answering the three research questions above, we plan to develop methods for estimating and calibrating the SA of human and robot agents in a multi-agent team through effective communication strategies, in the context of a specific task that the team is executing. We will develop methods that can relate the SA of the agents to the overall task performance, and will determine whether there exists an desired level of SA that leads to the best possible task performance by the team, under the circumstances.

#2.17