Hybrid Eyes: Design and Evaluation of the Prediction-Level Cooperative Driving with a Real-World Automated Driving System
archive: archived pipeline: cataloged verified
Get this paper ↗ (DOI — opens at the source; we link to it, we don't host it)
Summary
This paper addresses the limitations of current Automated Driving Systems (ADS) in predicting the behavior of surrounding traffic, a task where human drivers often outperform algorithms. To bridge this gap, the authors propose "prediction-level cooperation," a concept allowing human drivers to intervene by providing predictive insights about potential hazards. The study introduces PreCoAD (Prediction-level Cooperative Automated Driving system), an interactive prototype that integrates this concept into a real-world ADS algorithm called intelligent Traffic Flow Assistant (iTFA). The primary motivation was to move beyond previous Wizard-of-Oz studies by testing the concept with a functional ADS and improving system transparency through visual feedback. The researchers implemented PreCoAD in a driving simulator using IPG CarMaker. The system utilized a gaze-tracking eye tracker and a modified steering wheel with a piezoelectric sensor to detect double-taps. Drivers could select vehicles by looking at them and tapping the wheel, thereby "injecting" a prediction of potential hazard behavior (e.g., a cut-in) into the ADS. The system visualized its perception, planning, and predictions via a Graphical User Interface (GUI) to enhance transparency. A within-subject study was conducted with 15 participants across three sessions: baseline, experimental (with PreCoAD), and a second baseline. Participants drove five highway scenarios designed to challenge the ADS, including cut-in variations, tailgating, merge-ins, double lane changes, and erratic driving. The results demonstrated that PreCoAD enhanced automated driving performance and provided a positive user experience. Objectively, the experimental session showed a significant increase in Time-to-Passing (TTP), indicating the system allowed more time for safe maneuvering. Additionally, the number of situations with a headway below two seconds decreased significantly from a mean of 7.0 in the baseline to 4.8 in the experimental condition, and the total duration of such close-proximity driving was reduced. Subjectively, user experience scores were rated as "good" across all sessions, with the experimental session receiving slightly higher ratings. However, trust metrics showed no statistically significant differences between conditions. The significance of this work lies in validating prediction-level cooperation with a real ADS algorithm, confirming its utility in improving safety margins and driving comfort. Follow-up interviews revealed that while participants found the concept valuable, they emphasized the need for greater transparency regarding the system’s reasoning process. Specifically, users reported difficulty understanding when their input influenced the system and noted that visual indicators in the GUI were sometimes too small or unclear. These findings suggest that while prediction-level intervention is effective, future designs must prioritize clearer visualization of the ADS’s internal state and decision-making logic to fully leverage human-AI cooperation.
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.
| Stage | Outcome | Tool | Model | Prompt | Attempts | Completed |
|---|---|---|---|---|---|---|
| discover | success | Crossref | — | — | 1 | 2026-08-09 |
| archive | success | unpaywall | — | — | 2 | 2026-08-09 |
| extract | success | cached | — | — | 3 | 2026-08-10 |
| clean | success | clean | — | — | 1 | 2026-08-09 |
| chunk | success | chunk | — | — | 1 | 2026-08-09 |
| embed | success | embed | Qwen/Qwen3-Embedding-8B | — | 1 | 2026-08-09 |
| enrich | success | semantic_scholar | — | — | 1 | 2026-08-09 |
| promote | success | — | — | — | 1 | 2026-08-09 |
| summarize | success | llm | qwen3.6-27b-nvidia | summ-v5 | 2 | 2026-08-10 |
| tag | success | vector_similarity | — | — | 10 | 2026-08-11 |
| verify | success | — | — | — | 2 | 2026-08-10 |
Summary generated by qwen3.6-27b-nvidia on 2026-08-10; verification: verified.
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).
- Methodological Resource: tool software
- Theoretical Contribution: computational model, conceptual framework