The End of Reckless Driving
Can Autonomous Cars Reduce Traffic Accidents?

Akinori Morimoto
Professor, Faculty of Science and Engineering, Waseda University

The Current State of Traffic Accidents in Japan

Do you know how many people die every year due to traffic accidents? When traffic-related deaths reached its peak in 1970, the total number of deaths was 16,765 people (number of people who died within 24 hours of the accident). This statistic is comparable to the aftermath of the 2011 Tohoku earthquake and tsunami, which resulted in 15,894 deaths (as of February 2016). This means that during that time, there were mass traffic-related fatalities on the scale of the 2011 Tohoku earthquake and tsunami every year. In the years that followed, improvements in road conditions, the spread of safety education, the advancement of automobile technology and various other measures reduced fatalities to 3,904 deaths in 2016, going below the 4,000-mark for the first time in 67 years. Although continuous efforts have been made to create safer traffic conditions, the fact that many lives are still lost today due to traffic accidents is a reality we cannot deny.

The Current State of Traffic Violations

Why are there so many fatalities due to traffic-related accidents? There reasons are many, but out of the 3,790 fatal traffic accidents in 2016, 90 percent (3,410 accidents) of these accidents are attributed to traffic law violations (Table 1 shows the breakdown of these numbers). In other words, many of these accidents could have been prevented if traffic laws were not violated. If the advancement of autonomous vehicle technology paves the way for the complete automation of driving, it is expected that traffic violations will cease to exist, and a great majority of accidents will be prevented. One statistic in particular shows that a majority of fatal traffic accidents that occurred while the victim was driving (total of 1,338 deaths) could have been prevented with safer driving.

Table 1: Breakdown of fatal accidents due to traffic law violations in 2016
Law violated Number of accidents Law violated Number of accidents
Ignoring traffic lights 119 Drunk driving 24
Passage violation 192 Fatigue driving 33
Speed limit violation 199 Negligent driving 1,963
Right-of-way accident 104 Other violations 332
Pedestrian-related accident 252 Unknown violations 91
Failure to stop 101 Sum of all traffic violations 3,410

Source: National Police Agency website

Autonomous Driving Levels and Traffic Accidents

There are various definitions for the term "autonomous driving." By comparing the classifications defined by the U.S.-based SAE (Society of Automotive Engineers) in 2016 (see Table 2) and the rate of traffic accidents, it can be said that even by commercializing up to Level 2 on driving safety support systems can reduce the risks of various traffic accidents. Technologies such as automated distance control and collision avoidance systems can help avoid accidents or reduce the damage caused by them. Furthermore, Level 3 or above in autonomous driving technology can prevent some of the violations drivers commit on a regular basis, such as ignoring traffic lights and speeding, making the roads even safer.

Table 2: Autonomous driving level summary table (SAE J3016)
SAE Level Summary Main driving control
Level 0 (No Automation) The human driver controls all driving tasks. Driver
Level 1 (Driver Assistance) The system controls a specific function (such as steering or accelerating). Driver
Level 2 (Partial Automation) The system controls both steering and acceleration/deceleration. Driver
Level 3 (Conditional Automation) The system can drive automatically, but may require the human driver's attention and response. Automatic (may require driver intervention)
Level 4 (High-level Automation) The system is fully autonomous within the operational design domain. Automatic
Level 5 (Full Automation) The system is fully autonomous in all situations. Automatic

Source: Public-Private ITS Initiative/Roadmaps 2017

Risks of Implementing Autonomous Cars

Widespread use of autonomous cars (Level 3 or above) can pose new dangers when they are introduced on the same roads as conventional cars. For instance, even if the speed limit of a particular road is 50 km per hour, it is not uncommon for regular cars to be driving at speeds exceeding this limit. These drivers would of course be violating traffic laws, but the police do not have the capacity to stop all traffic violations, which has led to the normalization of speed limit violations. Figure 1 shows the percentage of enforced speed limit violations out of all enforced violations by region. The figure demonstrates how big of a problem speed limit violations are in certain regions.

Meanwhile, autonomous cars are system-controlled, meaning they strictly follow road traffic laws. This means that cars traveling at varying speeds are driving on the same roads, raising concerns of temporarily heightened traffic risks. Furthermore, autonomous cars follow traffic regulations by slowing down at yellow lights. Human-driven cars driving behind autonomous cars may try to speed up upon seeing yellow lights, heightening the risk of collisions.

Vision Zero—Building a Society Without Traffic Accidents

Unfortunately, commercializing the use of automated cars will not prevent all traffic accident-related fatalities. For example, by looking at the statistic of pedestrian deaths caused by traffic accidents in 2016, we see that 815 deaths out of a total of 1,324 (62 percent) were due to law violations, while 394 deaths of cyclists out of a total of 504 (78 percent) were also due to law violations. In order to prevent traffic accidents, not only automobiles, but other participants of traffic such as pedestrians and cyclists, must also follow traffic rules.

Roads would be comprised not only of system-controlled automobiles, but also non-system integrated participants, such as pedestrians and cyclists. In order to achieve Vision Zero—or a society without traffic accidents—in the face of technological advancements in autonomous driving, each individual must also be responsible for following traffic norms. It is also important to consider the method in which we introduce autonomous cars into our society. In order to adapt to new technology, societies and cities must also prepare to be held to a new standard.

The Three Implementation Policies with Regard to Introducing Autonomous Driving

We should carefully introduce autonomous driving while reducing the number of traffic accidents. To achieve this, I propose the following three policies for introducing autonomous cars (Level 3 or above) into our urban areas.

  • Start at low speeds
    Even if autonomous cars can flawlessly handle traffic conditions, it cannot completely prevent accidents caused by various factors such as pedestrians suddenly jumping into traffic. The risk of pedestrians suffering fatal injuries due to car collisions is only 10 percent when the vehicle is moving at 30 km per hour, but when the speed increases to 50 km per hour, the fatality rate jumps to 80 percent. To stress the importance of safety, it would be ideal to first start introducing autonomous cars that drive at low speeds.
  • Start in areas with low population density
    In central urban areas where human activity is rife—such as large numbers of pedestrians, cyclists, and transport vehicles, as well as risks including illegal parking—there is a relatively high rate of unpredictability from the perspective of autonomous cars. Considering this, we should start introducing autonomous cars in areas with low population density where there are fewer risks. This is especially ideal considering the fact that public transport is less accessible in areas with low population density, making the use of autonomous cars more socially practical.
  • Start from a low hardle fields
    It might be tempting to welcome a rapid introduction from a market competition perspective, but from an urban planning standpoint, it would be crucial to consider the balance of urban infrastructure and public awareness, and to implement a more gradual process. By slowly and gradually adjusting traffic infrastructure in areas with fewer obstacles, a more balanced approach to societal change can be expected.

Akinori Morimoto
Professor, Faculty of Science and Engineering, Waseda University

Akinori Morimoto completed graduate school at Waseda University in 1989. He earned a Ph.D. in engineering and is a Professional Engineer(Urban&Regional Planning).
Prior to taking his current post in April 2014, he had worked as a research associate at Waseda University, a research associate, associate professor and professor at Utsunomiya University, a visiting scholar at MIT (Massachusetts Institute of Technology), and other positions.
He specializes in transportation planning and city planning.
He has held official positions as a managing director at the City Planning Institute of Japan, managing director at the Japan Research Center for Transport Policy, and other positions.