Georgia Institute of Technology
Proactivity and Personalization in Longitudinal Robotic Assistance
Abstract
dc:description.abstractRobots are evolving toward versatile agents capable of assisting humans over extended periods. For robots to serve as effective general-purpose assistants, they must seamlessly integrate into users’ lives and proactively attend to their needs across long-term interactions. To function as truly proactive assistants, robotic systems must anticipate user behavior to provide timely assistance without needing to be asked, personalize the mode of assistance and manner of task execution to align with users’ preferences, adapt to the dynamic nature of human environments and the variability of tasks people perform, and interact with users to minimize errors, acquire missing information, and provide explanations. An ability to act proactively is fundamental to enabling robots to transition from command-following tools to capable assistants that can share responsibility for accomplishing tasks in everyday life. This dissertation presents a framework for proactive robot assistance and contributes computational meth- ods that enable robotic systems to build and leverage user-specific context to support proactive behaviors over extended interactions. Toward this goal, we investigate the following hypothesis: Anticipation of user be- havior and personalization to user preferences, across a wide range of tasks in dynamic environments, are key to proactive robot assistance over extended interactions. In support of this hypothesis, this disser- tation makes the following contributions: (1) a unifying formalism for proactivity in robot assistance; (2) a practical approach and dataset for longitudinal unprompted proactive assistance; (3) two computational meth- ods for anticipatory object movement prediction from unobtrusive user observations; (4) an algorithm to learn user preferences regarding robot assistance across a wide range of tasks from limited feedback; (5) an algo- rithm for a proactive robot assistant that asks informative questions and adapts task execution accordingly, along with a benchmark to evaluate it; and (6) an in-the-wild study of user perceptions of and interactions with an autonomous proactive robot assistant.
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
-
- Patel, Maithili
- Advisor dc:contributor.advisor
-
- Chernova, Sonia
- Committee members dc:contributor.committeemember
-
- Gombolay, Matthew
- Riedl, Mark
- Martín-Martín, Roberto
- Andreas, Jacob
- ORCID person:identifier.orcid
Rights
- Language dc:language.iso
- English
Identifiers
dc:identifier.*- Handle dc:identifier.uri
- https://hdl.handle.net/1853/82147