Unified composition of human behavior models for HRI

Develop unified composition methods for models of human behavior, motion, belief, desire, and intention that robotics developers can integrate into robotic systems to enable reliable prediction and interaction in human-robot environments.

Background

Human-robot interaction is critical for robots operating in daily-life tasks and mixed human-robot environments. While various models of human motion, beliefs, desires, and intentions exist, robotics development requires coherent integration and composition of these models to support trustworthy and predictable interactions.

The paper explicitly identifies the absence of unified composition mechanisms for these models as an unsolved problem, impeding developers' ability to build systems that can adequately model and predict human actions and reactions.

References

Although models of human behavior, motion, belief, desire, and intention are available, unifying and composing these models for use by robotics developers remains an unsolved problem.

— Software Engineering for Robotics: Future Research Directions; Report from the 2023 Workshop on Software Engineering for Robotics  (2401.12317 - Goues et al., 2024) in Challenges – Human Interaction and other Emergent Dynamic Behavior

While the capabilities for detecting and tracking pedestrians have improved, it is an open problem how to plan actions while factoring in probable human responses.

— A Human-Like Pedestrian Model for Automated Driving Simulations  (2609.29175 - Wang et al., 24 Sep 2026) in Section 1, Introduction

Developing a unified model with these capabilities therefore remains an important open problem.

— A Human-Like Pedestrian Model for Automated Driving Simulations  (2609.29175 - Wang et al., 24 Sep 2026) in Section 1, Introduction

Yet, how to integrate tactile feedback into learned handover policies remains an open question.

— Temporal Tactile Encoding and Compliance for Intent-Aware Robot-to-Human Bimanual Handover  (2609.05282 - Marra et al., 4 Sep 2026) in Section Introduction

Enabling models to reason beyond the visible, particularly regarding implicit human activities, remains an unresolved problem that requires advances in causal modeling and abductive inference.

— Commonsense Reasoning in Computer Vision: Foundations, Recent Advancements, and Future Directions  (2609.05257 - Mahmud et al., 4 Sep 2026) in Section 8.1, Reasoning Beyond Observables