Research Analyzer
← Back ICRA 2026

Responsibility and Engagement - Evaluating Interactions in Social Robot Navigation

Malte Probst, Raphael Wenzel, Monica Dasi

PDF

AI summary

New Responsibility and Engagement metrics accurately quantify how much each agent contributes to escalating or resolving conflicts in social robot navigation.
Social Robot Navigation Conflict Metrics Responsibility Engagement Trajectory Evaluation Human-Robot Interaction

Problem

Existing evaluation metrics for social robot navigation fail to capture the causal contribution of agents to conflict resolution or escalation. This gap hinders the assessment of behavior quality and the optimization of socially compliant planning algorithms.

Approach

The authors extend a prior Responsibility metric by adding time normalization to model the conflict buildup phase, and introduce an Engagement metric to quantify conflict intensification. They validate these metrics across simulated dyadic, group, and crowd navigation scenarios.

Key results

  • Extended Responsibility metric captures conflict buildup via time normalization
  • Introduced Engagement metric to quantify agent-driven conflict escalation
  • Metrics accurately attribute responsibility shares across dyadic, group, and crowd interactions
  • Validated for assessing navigation algorithm quality and foresightedness

Why it matters

Enables researchers and developers to quantitatively evaluate and optimize socially compliant robot navigation behaviors in complex, crowded environments.

Abstract

In Social Robot Navigation (SRN), the availability of meaningful metrics is crucial for evaluating trajectories from human-robot interactions. In the SRN context, such interactions often relate to resolving conflicts between two or more agents. Correspondingly, the shares to which agents contribute to the resolution of such conflicts are important. This paper builds on recent work, which proposed a Responsibility metric capturing such shares. We extend this framework in two directions: First, we model the conflict buildup phase by introducing a time normalization. Second, we propose the related Engagement metric, which captures how the agents’ actions intensify a conflict. In a comprehensive series of simulated scenarios with dyadic, group and crowd interactions, we show that the metrics carry meaningful information about the cooperative resolution of conflicts in interactions. They can be used to assess behavior quality and foresightedness. We extensively discuss applicability, design choices and limitations of the proposed metrics.

Index terms

Performance Evaluation and Benchmarking Human-Aware Motion Planning Motion and Path Planning

Related papers