The ability to make inferences about other’s mental states is referred to as having a Theory of Mind (ToM). Such ability is fundamental for human social activities such as empathy, teamwork, and communication. As intelligent agents being involved in diverse human-agent teams, they are also expected to be socially intelligent to become effective teammates. In this paper, we propose a computational ToM model which observes team behaviors and infer their mental states in a simulated search and rescue task. The model structure consists of a transformer-based language module and an RNN-based sequential mental state module in order to capture both team communication and behaviors for the ToM inference. To provide a feasible baseline for our ToM model, we present the same inference task to human observers recruited from Amazon MTurk. Results show that our proposed computational model achieves a comparable performance with human observers in the ToM inference task.
CITATION STYLE
Li, H., Le, L., Chis, M., Zheng, K., Hughes, D., Lewis, M., & Sycara, K. (2022). Sequential Theory of Mind Modeling in Team Search and Rescue Tasks. In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) (Vol. 13775 LNCS, pp. 158–172). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-031-21671-8_10
Mendeley helps you to discover research relevant for your work.