Human-AI Agent Interaction Design towards Agentic Cockpits for Intelligent Vehicles: Framework and A Case Study

This is the authors' abstract. We don't add key points for this paper.

The intelligent cockpit is evolving toward more human-like, agentic interaction. However, human–AI interaction in cockpits still faces challenges of fragmentation and unnaturalness. This study proposed a human-like interaction framework for cockpit AI agents. By mapping and transferring findings from in-depth interviews with human assistants, this study constructed a capability taxonomy and development path for cockpit AI agents, grounded in the abilities of LLM-based agents. Furthermore, through a case study of the meeting agent, this study illustrated how human–AI interaction could be enhanced across different stages. These findings provide insights for the development of agentic cockpits.