Export
Nathaniel S. Dennler
· 3 previous affiliations
- Massachusetts Institute of TechnologySince 2025
- Postdoc · 2025–2027
- Current
- University of Southern California2019–2024
- PhD student · 2019–2024
- Worcester Polytechnic Institute2015–2019
- Undergrad student · 2015–2019
Research Overview
Nathaniel S. Dennler focuses on human-robot interaction, socially interactive robot design, and robot-assisted clinical assessment. His work investigates how users perceive and interact with robotic systems, ranging from understanding the physical design metaphors of social robots to developing quantitative metrics for poststroke motor recovery. Recently, he has emphasized leveraging natural human behaviors to improve preference elicitation and align robot actions with user expectations.
Research Trajectory
The available evidence captures a focused snapshot of Nathaniel S. Dennler's research in 2025, demonstrating a multidisciplinary approach to human-robot interaction. As a lead author across distinct subfields, his work spans clinical assessment methodologies, algorithmic preference learning, and the psychological impact of robot embodiment.
In the domain of clinical robotics and rehabilitation, Dennler has investigated the phenomenon of "arm nonuse" in survivors of hemispheric stroke—a condition where patients overrely on their less-affected limb despite having recovered capacity in their paretic limb. In A metric for characterizing the arm nonuse workspacein poststroke individuals using a robot arm, he and his co-authors addressed the difficulty of obtaining ecologically valid evaluations of spontaneous arm choice. To quantify this gap between physical capacity and actual use, the research introduced the bimanual arm reaching test with a robot (BARTR), providing a structured metric for assessing nonuse without being skewed by patient awareness or presumed expectations.
Parallel to his clinical work, Dennler has advanced the computational mechanisms by which robots learn from human users. A central challenge in human-robot interaction is specifying the features of a robot's behavior that matter to users without requiring tedious manual labeling. In Contrastive Learning from Exploratory Actions: Leveraging Natural Interactions for Preference Elicitation, Dennler led the development of a framework that learns reward functions by observing users' natural exploratory actions. By applying contrastive learning to these interactions, the work seeks to extract semantically meaningful representations of behavior, thereby reducing the time and effort required for users to teach robots their preferences.
Dennler's research also extends to the physical design and social perception of robotic systems. Recognizing that a robot's physical embodiment heavily influences user expectations, he explored how design metaphors dictate the anticipated functionality of socially interactive robots. In Design Metaphors for Understanding User Expectations of Socially Interactive Robot Embodiments, Dennler contributed the Metaphors for Understanding Functional and Social Anticipated Affordances dataset. Based on crowdsourced user studies evaluating 165 extant robots, this work maps the alignment between a robot's physical appearance and its true capabilities, aiming to improve user adoption by ensuring that design cues accurately communicate functionality.
Current Directions
Based on the most recent papers in the evidence, Dennler's current research directions center on bridging the gap between human expectations and robotic capabilities. This encompasses both the physical design of robots—ensuring that their embodiment accurately signals their social and functional affordances—and the algorithmic frameworks necessary for robots to seamlessly adapt to user preferences. A recurring theme across his recent work is the desire to make human-robot interaction more natural and less burdensome, whether by utilizing spontaneous exploratory actions for preference elicitation or by developing unobtrusive, ecologically valid metrics for clinical rehabilitation.
Profile updated
| Title | Venue | Year |
|---|---|---|
| A metric for characterizing the arm nonuse workspacein poststroke individuals using a robot arm | Science Robotics | 2025 |
| Design Metaphors for Understanding User Expectations of Socially Interactive Robot Embodiments | Openreview | 2025 |
Loading new papers…
Other researchers whose papers cluster into the same research directions — a proxy for overlapping technical interests and shared intellectual neighborhoods.
See something wrong? Report a mistake