Comparison of Visual Sampling Patterns Under Simulated L2 and L0 Systems
Bibliographic record
Abstract
Automated driving systems (ADS) partially or fully perform or assist with primary driving functions. According to SAE J3016 (SAE, 2016), ADS can subsume driving tasks traditionally reserved for humans, ranging from L0 (no automation) to L5 (full automation), creating varying degrees of driver interaction and responsibility. However, the literature on human-automation interaction indicates that human operators may perform at a suboptimal level when interacting with automated support systems (Parasuraman & Riley, 1997), reducing the net benefit that automation can bring while also simultaneously increasing the potential for unforeseen human errors. Yamani and Horrey (in press) proposed a theoretical framework of human-automation interaction building upon a human information-processing model (Wickens, Hollands, Banbury, & Parasuraman, 2013) that accounts for human performance when interacting with varying types and levels of automation (Parasuraman, Sheridan, & Wickens, 2000). Following the model by Yamani and Horrey (in press), we hypothesized that when the ADS is perceived to be reliable, drivers engaging with such systems (e.g. L2) would exhibit eye movements no better or worse than the drivers engaged with manual or L0 driving since the drivers allocate their reserved or spare resources to other driving-irrelevant activities such as mind wandering or task irrelevant thoughts (Yanko & Spalek, 2014). The current driving simulator study compared young drivers’ eye movements across four unique scenarios in either L0 or L2 driving systems. We asked participants to complete a three-phased skill-based training program (RAPT-3; see Unverricht, Samuel, & Yamani for review) proven effective to improve young drivers’ ability to anticipate latent hazards, immediately followed by the evaluation of their eye movements in either L0 or L2 systems using a head-mounted eye tracker and a driving simulator. Participants in the L2 condition were instructed that the system detects and mitigates existing and latent threats on the forward roadway while maintaining appropriate speed and lateral positioning for the duration of the drive. To ensure similarity between both systems, L2 participants were required to position their hands on the steering wheel and feet above the pedal. No hazards materialized in any of the four driving scenarios. Data showed similar breadths of eye movements for the drivers of the L2 and L0 systems both horizontally [M = 36.5 vs. 36.3 pixels; L2 and L0, respectively] and vertically [M = 26.9 vs. 34.5 pixels] and no difference in mean fixation durations [M = 367 vs. 333 ms for L2 and L0 conditions]. However, data indicated substantial differences between L0 and L2 conditions for number of fixations, with L2 drivers fixating less frequently than L0 drivers, [M = 687 vs. 796 fixations, t (22) = 2.53, B 10 = 3.23]. The results imply that L2 drivers may sample information from the forward roadway less often than L0 drivers, suggesting the mobilization of spare resources for non-driving related tasks. Future research should examine the relationship between conveyed system reliability and attention allocation for drivers of ADS with different automation levels. In summary, the current results support Yamani and Horrey’s model and offer potential implications for the design of autonomous systems and the NHTSA automation guidelines to consider the perceived reliability of lower level ADS towards ascribing the role of the driver when the driving task is either partially or fully automated.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.004 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.001 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.001 | 0.000 |
| Open science | 0.000 | 0.001 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".