Georgia Institute of Technology
Methods and Models for Robots to Support Variable Human Task Performance
Abstract
dc:description.abstractRobots are increasingly deployed to assist users in various shared workspaces, including safety-critical domains such as defense, healthcare, and search-and-rescue missions, where performance standards are exceptionally high. Human-robot teams hold tremendous potential to enhance task execution by leveraging the complementary strengths of humans and robots. However, team performance can degrade due to errors, miscalibrated trust, or failures in mutual understanding. Achieving safe and efficient collaboration requires both humans and robots to accurately assess each other’s capabilities and determine when and how to rely on one another. To address these challenges, my thesis aims to identify key factors influencing human reliance on robots and develop computational techniques that empower robots to model users, adapt assistance, and share control effectively. I first investigate how robot behavior, user traits, and task context jointly influence users’ reliance on robots in sequential decision-making tasks. The findings reveal that adapting robot behavior can mitigate inappropriate reliance and that user expertise and task structure influence how users rely on robots, underscoring the need for modeling user behavior to support collaboration. Building on this foundation, I develop data-driven and model-based approaches that allow robots to infer user behavior and adapt in real time to improve performance. In this thesis, I examine robotic teammates that model user behavior across multiple roles: as passive observers, as partners with explicit divisions of labor, as reactive collaborators that intervene dynamically, and as mutual collaborators engaged in joint decision-making. In my final contribution, I address the fundamental question of when robots must infer teammates’ strategies in shared-task settings to succeed, as a function of task characteristics. Collectively, my work contributes novel algorithms and empirical insights for designing intelligent, adaptive robot teammates. Through human-subjects studies and extensive simulation-based evaluations, I demonstrate that the proposed algorithms enable robots to support appropriate reliance, infer user behavior, and determine how and when to adapt, resulting in significant gains in team performance.
Degree
thesis:*- Name thesis:degree_name
- Robotics, PhD
- Department thesis:degree_department
- Interactive Computing
- Grantor
- Georgia Institute of Technology
- Year dc:date.issued
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Natarajan, Manisha
- Advisor dc:contributor.advisor
-
- Gombolay, Matthew
- Committee members dc:contributor.committeemember
-
- Chernova, Sonia
- Feigh, Karen
- Admoni, Henny
- Riek, Laurel
- ORCID person:identifier.orcid
Rights
- Language dc:language.iso
- English
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/1853/82213