Research demonstrates that the proactivity of in-vehicle conversational assistants (IVCAs) can help to reduce distractions and enhance driving safety, better meeting users' cognitive needs. However, existing IVCAs struggle with user intent recognition and context awareness, which leads to suboptimal proactive interactions. Large language models (LLMs) have shown potential for generalizing to various tasks with prompts, but their application in IVCAs and exploration of proactive interaction remain under-explored. These raise questions about how LLMs improve proactive interactions for IVCAs and influence user perception. To investigate these questions systematically, we establish a framework with five proactivity levels across two dimensions—assumption and autonomy—for IVCAs. According to the framework, we propose a ``Rewrite + ReAct + Reflect'' strategy, aiming to empower LLMs to fulfill the specific demands of each proactivity level when interacting with users. Both feasibility and subjective experiments are conducted. The LLM outperforms the state-of-the-art model in success rate and achieves satisfactory results for each proactivity level. Subjective experiments with 40 participants validate the effectiveness of our framework and show the proactive level with strong assumptions and user confirmation is most appropriate.

Research demonstrates that the proactivity of in-vehicle conversational assistants (IVCAs) can help to reduce distractions and enhance driving safety, better meeting users' cognitive needs. However, existing IVCAs struggle with user intent recognition and context awareness, which leads to suboptimal proactive interactions. Large language models (LLMs) have shown potential for generalizing to various tasks with prompts, but their application in IVCAs and exploration of proactive interaction remain under-explored. These raise questions about how LLMs improve proactive interactions for IVCAs and influence user perception. To investigate these questions systematically, we establish a framework with five proactivity levels across two dimensions-assumption and autonomy-for IVCAs. According to the framework, we propose a "Rewrite + ReAct + Reflect" strategy, aiming to empower LLMs to fulfill the specific demands of each proactivity level when interacting with users. Both feasibility and subjective experiments are conducted. The LLM outperforms the state-of-the-art model in success rate and achieves satisfactory results for each proactivity level. Subjective experiments with 40 participants validate the effectiveness of our framework and show the proactive level with strong assumptions and user confirmation is most appropriate.

Material and texture decisions in bionic machinery installation art often remain intuition-dependent, limiting the reusability of empirical evidence for experience design. Building on the biomimetic content logic in biomimetic design theory, this study proposes a targeted framework—Texture Bionics—and operationalizes texture into four quantifiable perceptual dimensions: transparency, hardness, roughness, and surface texture, forming a controllable sample space of 12 plastic texture conditions. A case database encompassing 56 representative works (2000–present) was constructed to justify material selection; plastics were chosen for their tunable properties and feasibility for parameterized modulation. In a standardized viewing setup (≈500 lx illumination; 60 cm viewing distance), participants viewed a dynamic biomimetic mechanical wing module with interchangeable textured plastic surfaces. Subjective affect responses were captured using PAD ratings, and objective attention was assessed via wearable eye-tracking technology. Repeated-measures analyses showed robust main effects of texture on total fixation duration across all four dimensions, and selective effects on time to first fixation (significant/marginal for transparency, roughness, and surface texture but not hardness); pupillary response metrics provided no stable discrimination. PAD mappings further revealed functional “role types” (e.g., Key driver, Explore guide, Stable base), and a strong association between Arousal and inter-participant variability in fixation distribution, suggesting that high-arousal textures act as strategy amplifiers rather than uniformly increasing attention. Finally, findings were translated into an actionable Texture Design Toolkit using a three-question workflow—function label → attention goal → differentiation risk—to support evidence-based orchestration of installation narratives.