← All slides中文
01Research paper · Ligang Yan · April 2026

How autonomous driving could transform road safety

Most crashes are not caused by mechanical failure or bad roads. They are caused by people. This deck argues that a fully autonomous vehicle removes three specific human limitations, and shows the numbers behind each.

Originally written as a research paper for an academic English course; rebuilt here as slides. References on the last page.

02The problem

The weakest component in the transportation system is the driver

US road deaths, 2022
42,514
NHTSA, 2024
Serious crashes where the driver was the critical reason
~94%
NHTSA crash causation survey (Singh, 2015)
Road traffic crashes remain one of the leading causes of preventable death worldwide. The 94% figure has shaped a decade of debate about whether the human operator is the weakest part of the modern transportation system.
03Scope

What "autonomous" means here: SAE Level 5

SAE levelWho drives
0No automation; the human does everything
1–2Driver assistance; the human supervises constantly
3Conditional automation; the human must take over on request
4High automation within a defined domain (e.g. Waymo today)
5Full automation, all conditions, no human intervention
This paper's claim is about Level 5. No company, including Waymo and Tesla, has deployed fully autonomous vehicles at commercial scale. The argument is conditional: if they are successfully developed, three structural advantages follow.

Source: SAE International, J3016 (2021).
Advantage 1
Freedom from fatigue
Advantage 2
Consistent compliance with traffic rules
Advantage 3
Perception beyond the limits of human vision
04Advantage 1

A machine is not biological, so it cannot get tired

Commercial truck drivers judged fatigued in serious crashes
~13%
FMCSA Large Truck Crash Causation Study, 2006

Specific to trucking, and police-reported data are widely believed to understate fatigue across every category of driver.

  • Hours-of-service rules and mandatory electronic logging devices have not eliminated the risk. Fatigue is driven by biology, and laws cannot override biology.
  • An autonomous vehicle has no body clock and no sleep debt. It runs for as long as the power supply does.
  • Over 7.1 million rider-only miles across several cities, Waymo's deployed Level 4 system recorded no fatigue-related failures (Kusano et al., 2024).
05Advantage 2

It follows the rules, every time

Fatal crashes involving speeding
~29%
NHTSA, 2022
Alcohol-impaired driving
~32%
NHTSA, 2022
Distracted driving
~8%
NHTSA, 2022
Human drivers break the rules often, sometimes deliberately under time pressure and sometimes through misjudgement. Crashes usually have several causes, but deliberate or careless violations are the root of a large share of fatal ones.
  • An automated system operates inside a clear, bounded rule set. It will not speed or run a yellow light to save time, or change lanes without signalling, because legal alignment is designed in.
  • In Waymo's deployed fleet, Teoh and Kidd (2017) found no records of speeding, drunk driving or red-light running, behaviours that are routine among humans.
06Advantage 3

It sees more, farther, and faster than a human eye

Human limits
  • Glare. After oncoming headlights, vision takes about 0.8 s to recover for younger drivers and up to 2.1 s for older ones (Reuten et al., 2018). That gap contributes to night-time crashes.
  • Field of view. Combined horizontal vision approaches 200°, but only about 60° is sharp. Everything else needs a head turn or a mirror.
Waymo's sixth-generation sensor suite
Cameras
13
Lidar
4
Radar
6
Range
500 m
overlapping, all round, day and night

Waymo, 2024. The industry is split between multi-sensor fusion (Waymo) and vision-only (Tesla); which prevails is unsettled, but both far exceed human perception.

07Evidence from deployment

The advantages are already measurable

Waymo rider-only miles studied
7.1M
Multiple urban environments
Injury-reported crashes per million miles
0.41
Waymo Driver
Human benchmark, same measure
2.80
per million miles

An 85% reduction in injury-reported crashes, on a Level 4 system that still has limits a Level 5 system would not.

Kusano, Scanlon, Chen, McMurry, Chen, Gode and Victor (2024), Traffic Injury Prevention.

08Conclusion

Three limitations removed, three documented causes of death addressed

Biological
Not alive, so never fatigued.
Behavioural
Bound by rules, so far less likely to speed, drive impaired, or be distracted, the violations behind most fatalities.
Perceptual
Sensor arrays instead of eyes, so hazards are detected faster, farther, and in darkness and weather.
Important questions remain, above all which sensor architecture wins. But the structural case is clear: by removing the biological, behavioural and perceptual limits that make human drivers error-prone, autonomous vehicles offer a credible path to safer roads.
09References

Sources

· Federal Motor Carrier Safety Administration (2006). Report to Congress on the Large Truck Crash Causation Study. US DOT.
· Kusano, K. D., Scanlon, J. M., Chen, Y.-H., McMurry, T. L., Chen, R., Gode, T., & Victor, T. (2024). Comparison of Waymo rider-only crash data to human benchmarks at 7.1 million miles. Traffic Injury Prevention, 25(sup1), S66–S77.
· National Highway Traffic Safety Administration (2024). Overview of motor vehicle traffic crashes in 2022 (DOT HS 813 560).
· Reuten, A. J. C., et al. (2018). Headlight glare exposure and recovery in younger and older drivers.
· SAE International (2021). Taxonomy and definitions for terms related to driving automation systems for on-road motor vehicles (J3016_202104).
· Singh, S. (2015). Critical reasons for crashes investigated in the National Motor Vehicle Crash Causation Survey (DOT HS 812 115). NHTSA.
· Teoh, E. R., & Kidd, D. G. (2017). Rage against the machine? Google's self-driving cars versus human drivers. Journal of Safety Research, 63, 57–60.
· Waymo (2024). Meet the 6th-generation Waymo Driver. Waymo blog.