Trust Modeling from Sparse Signals: A State Space Approach for Human-Robot Interaction

Social robots are increasingly used to help humans with various tasks, from healthcare and education to emergency response. Their success depends heavily on trust, which is the belief that someone or something will act reliably, safely, and as expected. In human-robot interaction, trust is about how much a person believes the robot will behave correctly and support them effectively. In high-stakes scenarios or unfamiliar situations, such as fire emergencies, humans must rapidly decide whether to follow robot guidance. Trust in these contexts is dynamic, context-dependent, and high-stakes. Therefore, trust modeling is essential for safe and effective human-robot interaction.

Assignment

This project focuses on modeling trust as a dynamic state using a learning-based state space framework. The research question is: How can human trust be modeled and under what conditions is it identifiable from sparse behavioral signals?

Trust is treated as a hidden variable that evolves over time based on the robot's behavior, the environment, and human responses. In other words, trust, as a latent state, is not directly observable and should be inferred from sparse signals (i.e., limited behavioral cues/measurements). Observations, such as adherence to guidance, task performance, or situational feedback, sequentially update trust. By learning this model from data, the robot can estimate human trust in real time and anticipate human responses. Developing an accurate and adaptive trust model lays the foundations for sequential decision-making systems, enabling robots to plan actions under uncertainty while accounting for dynamic trust levels. 

The project will be jointly supervised by Anahita Jamshidnejad (A.Jamshidnejad@tudelft.nl) and Geethu Joseph (g.joseph@tudelft.nl).

Requirements

We are looking for a student from electrical engineering or control engineering with a strong mathematical background, e.g., with a BSc background in either electrical or mechanical engineering. Good linear algebra skills are required. 

Contact

Last modified: 2026-02-11