While vehicle safety has been dramatically enhanced over the past decades, driver error remains a top cause of car accidents. Fully autonomous vehicles would remove human error from the equation, and this single change could help reduce car accidents. While there’s been talk of self-driving vehicles being able to eliminate accidents entirely, that’s not quite possible. However, based on crash data and information, autonomous vehicles do stand posed to make the roads safer in many ways.
Driver Error and Car Accidents
The National Motor Vehicle Crash Causation Survey (NMVCCS) was conducted from 2005 to 2007 and shed light on the most common factors that led up to automotive crashes. The survey assessed a sample of 5,470 crashes and found that 94% of crashes were caused by driver error. Vehicle-related issues and failures, environmental conditions and unknown critical reasons each accounted for 2% of the crashes.
The survey further broke down the type of driver errors that resulted in crashes. Recognition errors were the most common, accounting for 44% of accidents, with decision errors and performance errors the second- and third-most-common causes.
Based on the survey data, driver error would account for 2,046,000 of the estimated 2,189,000 crashes nationwide over a two-and-a-half-year period.
With driver error causing the majority of accidents, fully autonomous vehicles would remove human error from the equation. Autonomous vehicle technology doesn’t get distracted, doesn’t feel pressure to speed to make up for running late and isn’t influenced by emotion, fatigue or boredom. Applying the NMVCCS data indicates that, should we reach a point where all vehicles on the road are Level 4 or 5 autonomous vehicles, most or all of that human error could be removed. Theoretically, 94% of vehicle crashes could be eliminated.
A recent study by the Insurance Institute for Highway Safety (IIHS) examined 5,000 police-reported crashes, reviewing each case for driver-related factors that contributed to the crash. The study broke down the causes into categories:
- Sensing and perceiving errors like distracted driving caused 23% of crashes.
- Incapacitation due to falling asleep or drugs and alcohol caused 10% of crashes.
- Planning and deciding errors like driving aggressively or too fast caused 40% of crashes.
Predicting errors like misjudging gaps in traffic and execution and performance errors like performing evasive maneuvers were also causes.
The study’s results highlight opportunities for autonomous vehicles to make roads safer. If all vehicles were fully autonomous, sensing and perceiving errors and incapacitation could be eliminated as crash causes, preventing 33% of accidents.
Even a 33% reduction in accidents would be significant and life-saving. The NHTSA statistical projection of traffic fatalities for 2019 indicates that approximately 36,120 people died in traffic accidents. A 33% reduction in accidents would, presumably, save more than 10,920 lives, confirming the safety potential that self-driving vehicles hold.
Further Designing Autonomous Vehicles for Safety
While autonomous vehicles would eliminate incapacitation and sensing and perceiving errors, other human causes could still be factors unless autonomous vehicle technology is built to prevent those issues. If human drivers retain the control and ability to speed or break laws in autonomous vehicles, then accident causes like planning and deciding errors and predicting will continue.
The IIHS study predicts that self-driving vehicles would only reduce human error-caused crashes by one-third, but that doesn’t mean that autonomous vehicles can’t help make the roads safer. Instead, it identifies the other parts of the crash equation that these vehicles will need to remedy.
How autonomous vehicles are programmed will determine how well they’re able to prevent other human error crash causes, including planning errors like driving too fast for road conditions, and executing incorrect evasive maneuvers. That programming responsibility lies with the automakers, and it also provides an opportunity to truly make these vehicles safer, while also making them an asset to society.
Autonomous vehicle programming determines how the vehicle will interact with and react to other cars, humans, animals and obstacles on the road. That programming also determines when the vehicle executes an evasive maneuver, and which type and degree of maneuver is suitable for a situation. The more extensive and detailed the programming, the more appropriately the vehicle will be able to react in an unforeseen situation – and the better its chance of preventing an accident.
Preparing vehicles for the unforeseen and unusual circumstances they may encounter takes time and road miles. Astro Teller, head of Alphabet X lab, notes that the self-driving vehicle project, Waymo, has wholeheartedly embraced that development and evaluation task. During an interview at the EmTech Digital conference in San Francisco, Teller explained that Waymo vehicles had driven 2.5 billion miles in simulations over the course of 12 months. The vehicles had also traveled more than five million miles on roads with drivers.
Waymo also presented the vehicles with “pathological situations” to test how the technology would respond in truly unpredictable events. Those situations included having people suddenly appear in front of the vehicles dressed as Elmo, while lying on skateboards and while hiding in bags.
While this detailed preparation can help equip vehicles to respond appropriately to unusual circumstances, vehicles will also need to be designed to follow traffic laws, even if their human passengers want to travel at unsafe speeds. Vehicles will need to respond to hazardous road conditions and adapt to higher-risk areas, such as those that are heavily populated with pedestrians.
The goal of designing self-driving vehicles might appear to be to make them drive like human drivers, but the key to accident prevention is actually to make these vehicles operate more like machines. By removing the ability to customize the vehicle operation to the rider’s preferences, manufacturers can make self-driving vehicles safer by preventing them from speeding. Autonomous vehicles have many advantages, like faster reaction times than humans. Designing them to operate with a machine-life safety prioritization could make them safer than human drivers.
Finally, autonomous vehicles offer the benefit of on-demand transportation and easy vehicle sharing. This can help to reduce the number of vehicles on the road, minimizing traffic congestion and decreasing single-passenger trips and unnecessary travel. These vehicles can identify and take the most efficient route available, spending less time on the road. Fewer vehicles and less traffic would also help to reduce the number of car accidents that occur.
There are still many hurdles to overcome in the development of fully autonomous vehicles. The technology, including the laser, radar and camera sensors that self-driving cars are equipped with, is steadily improving yet still needs refining and proven reliability in real-world situations. However, by analyzing and understanding the details of car accident causes, vehicle manufacturers can identify how these vehicles need to be designed to help prevent as many accidents as possible. Autonomous vehicles won’t just be safe additions to the roads, but could actually make those roads safer.