Using a user-independent approach for automotive human-machine interface assessment

Biondi, F; Cooper, JM; Strayer, DL · 2018 · publications_jsonl

DOI: 10.4135/9781526431585

archive: archived pipeline: cataloged

Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)

Summary

This paper addresses the inefficiency of traditional automotive human-machine interface (HMI) assessment methods, which are resource-intensive and often fail to identify specific design flaws responsible for high driver demand. Motivated by the rapid proliferation of complex in-vehicle infotainment systems, the authors propose a user-independent approach grounded in the visual search task paradigm. This method shifts the focus from measuring driver interaction outcomes to analyzing intrinsic interface characteristics, such as layout, menu structure, and system responsiveness, to generate usability scores without requiring human participants. The approach is designed for early-stage development and benchmarking, allowing manufacturers to pinpoint design causes of poor usability. The methodology involves evaluating six widely used in-vehicle systems: HondaLink, FCA UConnect 3, Ford Sync3, FCA UConnect 8.4, Hyundai’s proprietary system, and Chevrolet MyLink. The assessment focuses on two primary interface types: visual displays and voice-based systems. For visual interfaces, key metrics include menu depth (number of steps to access functions), system responsiveness (time from input to response), and proportional target button size relative to function frequency. Redundant buttons, which provide duplicate access to the same functions, are identified as a source of confusion. For voice interfaces, the analysis considers the number of interaction steps and the naturalness of required speech commands, drawing on psycholinguistic principles regarding semantic networks and speech production. Aggregate usability scores were calculated for each system based on these design elements. To validate this user-independent scoring system, the authors conducted a cross-validation on-road study with 32 participants. Participants drove vehicles equipped with the six systems and performed tasks such as tuning the radio to 99.5 FM and placing calls. Subjective workload was measured using the NASA-TLX and System Usability Scale. Results indicated significant variability in interaction complexity; voice-based tasks ranged from one to three steps, with average task times varying from 13 to 23 seconds. Visual interface responsiveness varied dramatically, ranging from 0.3 seconds to 4 seconds. Statistical analysis using linear regression demonstrated strong correlations between the calculated aggregate design scores and the subjective workload and usability ratings collected during the on-road study. This confirms that the user-independent metrics successfully predict driver demand and usability. The significance of this work lies in providing a scalable, cost-effective tool for automotive manufacturers to assess HMI usability during the design phase. By linking specific design elements—such as button size, menu depth, and system lag—to predicted user demand, the approach helps identify root causes of poor usability before expensive user testing is conducted. It bridges the gap between experimental psychology principles, such as visual search and semantic mapping, and applied automotive engineering, offering a framework for benchmarking new interfaces against established standards.

Key finding

Aggregate user-independent scores derived from visual-search-grounded interface features (menu depth, responsiveness, button layout, voice interaction steps) showed strong linear-regression correlations with on-road workload and usability ratings across six infotainment systems, supporting design-only assessment as an early-stage proxy for driver-in-the-loop testing.

Methodology

on_road

Sample size: Cross-validation: N=32 drivers across six infotainment systems

Provenance

The full processing record for this entry. Every stage of this paper's journey through the pipeline is logged — what ran, with which tool and model, how many attempts it took, and when it last completed. Discovered via tag_papers on 2026-05-30 (4 acquisition events logged).

StageOutcomeToolModelPromptAttemptsCompleted
discover success 1 2026-05-06
archive failed pmc 8 2026-06-04
extract success cached 12 2026-08-22
clean success clean 2 2026-08-10
chunk success chunk 2 2026-08-10
embed success embed Qwen/Qwen3-Embedding-8B 2 2026-08-10
enrich failed 5 2026-08-07
promote success 2 2026-06-06
summarize success llm qwen3.8-27b-gittensor summ-v5 2 2026-08-22
tag success vector_similarity 27 2026-08-11
verify success 1 2026-05-08

Summary generated by qwen3.8-27b-gittensor on 2026-08-22; verification: pending re-verification.

Topics

Ranked by relevance to this paper. Hover a topic for its definition.

Information type

What kind of knowledge this paper contributes, grouped by family — independent of topic (what it is about) and method (how it was studied).