In goal recognition problems, the goal is to identify the goal of an observed agent as quickly as possible before they reach their goal. This problem can be applied to a number of applications ranging from software personal assistants and robots that anticipate the needs of humans; intelligent tutoring systems that recognize sources of confusion or misunderstanding in students through their interactions with the system; and security applications that recognize terrorists plans.

In goal recognition design problems, the objective is to identify the best ways to modify the underlying environment of the agents in such a way that they are forced to reveal their goals as early as possible. This problem is typically applicable in the same goal recognition applications as long as the underlying environment can be modified. For example, in intelligent tutoring systems, it is possible to modify the sequence or type of questions asked so that students’ misunderstanding can be identified sooner. Similarly, in security applications, it may be possible to block some paths so that an attacker’s plan can be recognized earlier.

Our current research is on both the algorithmic back-end, where the goal is to enrich the models so as to better capture more realistic problem characteristics and develop efficient algorithms to solve them, as well as the user interface front-end, where the goal is to develop visualization interfaces that allow our methods to be adapted to specific applications and evaluated with human users.


Improving Client Experience Through Goal Recognition and Explainable Assistance in Adaptive Systems.
J.P. Morgan Chase Bank (2022 – 2023).

CAREER: Decentralized Constraint-based Optimization for Multi-Agent Planning and Coordination.
National Science Foundation (2016 – 2021).

Integrated Computational and Cognitive Workflows for Improved Security and Usability.
Boeing (2019).

Representative Publications