Building Trust in Multi-Robot Systems through Shared Mental Models

In this article, we explore how to build trust among stakeholders in complex robotic tasks involving multiple robots and parties. The authors argue that traditional methods of building trust through one-on-one interactions are not effective in such scenarios due to the high complexity of the environment. Instead, they propose leveraging shared mental models to update trust dynamics.
A shared mental model refers to a collective understanding of how a task or system works, which is constructed by observing and interacting with multiple robots and stakeholders. This model can help stakeholders align their expectations and perceptions of the task, leading to greater coherence in their mental models. By regularly sharing information and updating their mental models together, stakeholders can establish a more accurate and reliable understanding of each other’s perspectives.
The authors describe three key strategies for building trust through shared mental models: briefing and debriefing, elaboration and visualization, and cross-training. Briefing involves sharing information at the outset and scheduled points during the larger task, while debriefings provide a corrective adjustment for divergence in mental models over time. Elaboration and visualization involve using both augmented reality (AR) and virtual reality (VR) mediums to observe the state of the task and robots spatially, allowing stakeholders to access current information from their preferred perspective.
The article highlights the importance of evaluating subjective experiences during use by embedding questionnaires into VR, enabling a seamless evaluation of trust and shared mental models during the task. By leveraging shared mental models, stakeholders can establish more accurate and reliable trust dynamics in complex robotic tasks involving multiple parties.
In summary, this article demonstrates how shared mental models can help build trust among stakeholders in multi-stakeholder multi-robot tasks by providing a collective understanding of the task and its dynamics. By regularly sharing information and updating their mental models together, stakeholders can establish more accurate trust dynamics that take into account the complexity of the environment.