Waiting for the Big Breakthrough in Autonomy

The Potential to Reduce 90% of Road Fatalities

About 1.35 million people are killed worldwide in traffic accidents each year. That’s more fatalities than from AIDS, tuberculosis or even the most extreme projections of the coronavirus (COVID-19). Unsurprisingly 94% of all serious crashes were caused by driver-related factors, such as impaired driving, distractions, speeding or other illegal driving behavior.  

Fully autonomous vehicles (AVs), unlike human drivers, won’t get distracted, fatigued, or influenced by alcohol. According to the US Department of Transportation, the potential of autonomous vehicles to reduce injuries and deaths cannot be overstated.

In 2016, scientists launched the Moral Machine, an online survey that explored moral issues regarding autonomous vehicles in unavoidable accidents. Participants were asked which potential victims should be spared based on their age, gender and socio-economic background. In other words, should an autonomous car hit two children or two joggers and an old person? Interestingly, people from different cultures provided different answers. Other scientists believe that deliberating over which people to hit is like worrying about asteroid strikes. 

Source: Nature

The Five Levels of Autonomy

But to really understand the issues surrounding autonomous driving, it’s important to take a deeper look. The Society of Automotive Engineers has defined five levels of driving automation that range from Level 0, which includes driver support features such as automatic emergency braking, to Level 5, which is fully automatic/autonomous driving under all conditions.

Reducing accidents by up to 90% could save the US economy nearly $190 billion since for every person killed in a motor-vehicle accident, 8 are hospitalized and 100 are treated in emergency rooms.

In addition to the cost savings, autonomous vehicles could free up as much as 50 minutes a day for riders who will be able to work, relax, or access infotainment instead of driving. Globally, commuters may save up to one billion hours—about 114,000 years—and spend up $5.6 billion per year on digital media and services.

The Long (?) Road Ahead

A few years ago, most of the OEMs and AV vendors predicted that they would offer full, Level 5 autonomy by 2020 or 2021. Tesla is running ahead of schedule and may release its full autonomy Autopilot feature for a limited beta test by the end of 2020.

However, most OEMs and other industry players have admitted that the challenge has been a bit more formidable than previously thought. In April 2019, the CEO of Ford publicly stated that they had “overestimated the arrival of autonomous vehicles” and that Ford’s 2021 AVs would be geo-fenced and operate only within specific areas that have been precisely and carefully mapped.

So what happened? First, human driving behavior and the outliers—the unusual events—have been proven to be more erratic and more difficult to anticipate than previously thought. Second, the AV industry has been stung by 16 high-profile fatalities since May 2016 involving cars with Level 2 and Level 3 autonomy and drivers (or “safety drivers”) who failed to reassert control over the vehicles, usually because they were not following proper procedures, a fact that was omitted from most news coverage.

One of the most infamous  accidents occurred in March 2018 when an Uber self-driving vehicle killed a jaywalker. The accident was caused by a series of errors, ranging from a single safety driver who was allegedly watching a video to poor programming logic for predicting the movements of pedestrians who neglected to use crosswalks (a.k.a. law-breaking jaywalkers).

Steve Wozniak, who used to be an avid AV evangelist is now unsure whether Level 5 autonomy will occur in his lifetime. GuardKnox CEO, Moshe Shlisel, doubts whether people will entrust the wellbeing of loved ones to driverless Level 5 robo-taxis due to cybersecurity concerns and ethical issues surrounding unavoidable accidents.

In addition to the cybersecurity issues, the erratic behavior of human drivers, and recklessness of jaywalking pedestrians, autonomous vehicles will need to be 20-40 times faster than today... Vehicles with semi-autonomous features currently have network speeds ranging from 500 Kbps to 1 Mbps, but Level 4 and 5 autonomous vehicles will require network speeds of 10-20 Gbps in order to support . LIDAR and numerous other sensors that will generate

Worth the Wait

While full Level 5 autonomy that’s able to handle Bolivia’s “Road of Death” may never arrive, the ability to save more than a million lives each year is enough incentive to keep working.

In 2019, autonomous test vehicles from 36 companies drove 2.9 million miles in California and Nuro’s R1 vehicle started delivering groceries and pizzas in Arizona and Texas, even though its vehicles lack side-view mirrors or a windshield. In other words, incremental progress is being made, even on the regulatory front.

Source: Nuro

The 2020 Tokyo Olympics, now scheduled for 2021, were slated to be a big reveal of autonomous vehicle technology that would demonstrate Japan’s technological prowess. Plans for the Olympics include more than 3,700 mobility products and autonomous vehicles, ranging from Level 4 robo-taxis to Toyota’s e-Palette modular, multi-purpose autonomous vehicles.

The upside of the new date for the Olympics is that there will be more time to improve the performance of the AVs as well as implement the cybersecurity solutions that can mitigate the potential risks to AVs and connected vehicles from hackers.