Home / Insights / Riveting Self-Driving Car Statistics (2023)

Riveting Self-Driving Car Statistics (2023)

By Jason L Arthur
April 4, 2023
Fact Checked

While there are currently no fully driverless vehicles on the road, barring those being used in testing, it’s full steam ahead for the development of self-driving cars, or automated vehicles (AVs).

AVs have the potential to take stressed, tired, intoxicated, and distracted drivers off the road to bring about safer driving. But it isn’t just consumers who feel they can benefit from these futuristic vehicles. In fact, 55% of small business owners believe their fleets will be completely autonomous in the next two decades [11].

Still, public perception is mixed. Below, we’ll explore a myriad of self-driving car statistics so you can learn more and see where you stand on the topic.

Key Insights

There are currently more than 1,400 self-driving cars in the U.S. being tested by 80 companies [12]
There are more than 30 million automated vehicles (AVs) with at least level 1 autonomy worldwide [13]
In the U.S., the current AV market is estimated to be valued at around $4 billion [2]
34 states and Washington, D.C. have published legislation regarding AVs [20]
94% of car accidents are caused by human error [2]
An estimated 1 out of 3 crashes could be prevented by driverless cars [15]
The accident rate of self-driving cars is 9.1 crashes per million miles driven—more than double the rate of regular cars [8]
According to the NHTSA, there have been 419 AV crash reports and 18 fatalities [28]
The most common type of accident involving self-driving cars involves getting hit from the rear (62%) followed by sideswipes (21%) [18, 19]
12% of U.S. respondents feel safe in self-driving cars; however, 72% would feel safer if they had the ability to step in and take control in emergencies [10]
55% of small business owners believe their fleets will be completely autonomous in the next 20 years [11]
At best, up to 20% of new passenger cars will be sold with level 3 or higher autonomy by 2030 and 57% by 2035 [26]

What are self-driving cars?

AVs use technology to either partially or fully replace human drivers with Advanced Driving Assistance Systems (ADAS) [4, 6]. Virtually all new cars today have some level of driver assistance technology, like adaptive cruise control and automatic emergency braking, which are meant to assist the driver in reducing their workload [3]. However, none of these systems is currently reliable enough to allow drivers to take a complete back seat, so to speak [3].

To clarify how autonomous AVs truly are, the Society of Automotive Engineers (SAE) has developed a widely-accepted classification system based on the level of human intervention required [6]. See the full breakdown of each level below [1].

Level 0: Momentary driver assistance offers warnings and alerts, like lane departure warnings and automatic emergency braking, while the driver is fully engaged and attentive
Level 1: Driver assistance provides continuous assistance with either acceleration and braking OR steering, like lane keeping assistance or adaptive cruise control, while the driver is fully engaged and attentive
Level 2: Additional assistance provides both acceleration and braking AND steering, like highway pilot (similar to autopilot), while the driver is fully engaged and attentive
Level 3: Conditional automation in which the system actively drives, handling all aspects of driving, with the driver attentive and ready to take over
Level 4: High automation, when engaged, means the system drives and is fully responsible for all driving tasks. The human driver is a passenger and does not need to be engaged; however, these vehicles can only operate in limited service areas.
Level 5: Full automation, when engaged, means the system drives and is fully responsible for all driving tasks. The human driver is a passenger and does not need to be engaged; these vehicles can operate universally in all conditions and on all roadways.

Levels 0-2, which offer driver support features, are currently available in the U.S., with levels 3-5 still in development and not yet available for consumer purchase [1, 4].

How many self-driving cars are there?

It’s difficult to say how many self-driving cars there are in the U.S. and abroad. First, the definition of “self-driving” cars depends on the specific classification based on the SAE’s levels. Second, fully autonomous vehicles (levels 3-5) are not yet commercially available and are still in testing. Still, states allow a limited number of self-driving vehicles on designated public streets for testing, research, and pilot programs [1].

According to a recent talk at the Uber Elevate Summit, the U.S. Department of Transportation revealed that there are currently more than 1,400 self-driving cars in the U.S. being tested by 80 companies [12]. California was not only one of the first states to test AVs on public roads but also the state which holds the majority of these test companies—about 62 in total [12].

By contrast, it is estimated that there are more than 30 million AVs with at least level 1 autonomy worldwide [13]. While there is still no official data, it is also estimated that this number will reach over 54 million by 2024 [14].

Define “testing”

Let’s use Tesla as an example. Tesla vehicles are currently at level 2 autonomy [21]. Their fully self-driving vehicle is still in testing, with roughly 100,000 drivers currently testing the software [21]. After passing a safety test, these test drivers install the Full Self Driving (FSD) Beta software on their vehicles, running updates whenever available, to observe whether any negative behavior is an isolated event or if it’s rare or consistent [21].

In testing, these drivers are helping Tesla see what changes need to be made in order for FSD Beta to reach level 5 autonomy, meaning fully autonomous, requiring no human intervention [21]. Just what are they looking out for? The vehicle’s ability to select the correct lane when turning at an intersection, slowing down at speed bumps or dips in the road, phantom braking and stopping, and running stop signs, among other things [21].

All of this information provides the AV industry with valuable insights, helping to bring level 5 vehicles onto the road [21].

What kind of technology is involved?

Several companies, like NVIDIA and Advanced Micro Devices (AMD) are making chips for AVs. To understand the kind of technology involved, let’s look at NVIDIA as an example.

NVIDIA has an AV platform, NVIDIA DRIVE, which is a suite of products for AVs [22]. Through DRIVE, NVIDIA handles everything from the vehicle’s computer to sensor data collection to building models for perception, mapping, planning, and control [22]. Its newest artificial intelligence (AI) platform to power AVs was announced in 2021, the NVIDIA DRIVE Orin SoC (system-on-a-chip) [22]. This technology serves as the central computer for AVs and promises 254 TOPS (trillion operations per second) [22]. For comparison, Tesla’s FSD Computer operates at 144 trillion TOPS [22]. That’s seriously quick thinking!

Using this NVIDIA technology allows developers to not only build but also scale and leverage their developmental investment across autonomy level 2 to level 5 vehicles in their fleet [22].

Are self-driving cars safer?

Without enough data, it’s too soon to say for sure how much safer AVs truly are. However, with 40,000 car accident fatalities each year and 94% of car accidents being caused by human error, ADAS systems should, in theory, reduce car accidents by up to 94% [2]. There is no distracted or impaired driving in a self-driving car. Of course, like humans, technology can also be flawed. While ADAS systems won’t text and drive or drink and drive, below are some of the ways in which these systems need work [4]:

AVs perform worse in edge cases, which are essentially those extreme or incredibly unlikely situations
Bad weather can leave AV technology prone to error
Inner-city traffic can be difficult for AVs to navigate
Computers lack intuition and instinct, which can affect response time and more
Many safety regulations are not mandatory for AVs
It is difficult to standardize laws across states

The above points highlight some of the many reasons why levels 3-5 are still not available and require improvements and fine-tuning until they’re available for purchase.

Accidents and fatalities

Car accidents claim thousands of lives each year, but self-driving cars are not a perfect solution. In looking at the accident rate of self-driving cars, they account for 9.1 crashes per million miles driven—more than double the rate of regular cars at 4.1 crashes per million miles driven [8]. However, on a positive note, these crashes tend to result in far less severe injuries [4].

According to the NHTSA, there have been 419 AV crash reports, with 263 of those crashes involving level 2 AVs and 156 involving fully autonomous vehicles [28]. As a result, there have been 18 deaths, but all of those involved the former group, rather than the latter, fully autonomous group [28]. However, the NHTSA does not require manufacturers to report whether or not the crash was caused by a technical issue or driver error [28].

How many accidents self-driving cars may reduce

Again, the lack of data makes it hard to say just how many accidents self-driving cars can reduce. However, after analyzing 5,000 crashes, the Insurance Institute for Highway Safety (IIHS) concluded that one out of three crashes could be prevented by driverless cars [15]. The IIHS went on to say that of those crashes AVs prevented, they could likely reduce only those crashes caused by driver errors [15]. In response, the AV industry noted that AVs are programmed to prevent a much higher number of crashes—about 72%—caused by far more complex errors [15]. Some manufacturers, such as General Motors, say AVs are an essential part of eventually achieving a zero crash rate [15].

Regarding the types of accidents, the most common type of accident involving self-driving cars involves getting hit from the rear (62%) followed by sideswipes (21%) [18, 19]. These statistics help to show that most accidents involving AVs occur due to driver error on the other car’s part.

Public perception

Despite their potential for greater safety, AVs bring up a mixed bag of emotions. A recent poll revealed the following sentiments [9]:

2 in 10 Americans think AVs will never be safe
75% think AV technology isn’t ready for consumer use
Almost 50% said they wouldn’t get into an AV rideshare or taxi
34% believe that the pros of AVs outweigh the cons

In terms of safety, another study revealed that just 12% of U.S. respondents feel safe in self-driving cars; however, 72% would feel safer if they had the ability to step in and take control in emergencies [10].

The main reason many drivers feel unsafe in AVs is the false sense of security they provide by being a passenger, rather than a driver. As previously mentioned, these vehicles are not fully autonomous yet, but the varying levels of “self-driving” cars are not widely known to the public, creating a bit of a misnomer. Additionally, the risk of cyber-attacks poses a safety threat, as do the lithium-ion batteries used in AVs, which are volatile and combustible [10, 8].

Legislation

Lawmakers are struggling to agree on how best to regulate AVs, raising questions over safety and liability rules [25]. In 2018, the AV Start Act, which would have introduced a framework for testing, regulating, and deploying AVs, was brought to Congress but failed to pass [6]. Despite the AV industry’s push for federal regulation since then, the federal government has published voluntary guidelines for states to enact their own legislation [24].

Some states allow the deployment of AVs for commercial use only, while others allow AVs for testing only [20]. In total, 34 states and Washington, D.C. have published legislation regarding AVs, but not all of these laws align on key factors such as whether or not there needs to be a driver present, whether or not that driver must be licensed, and whether the vehicle requires liability insurance [20]. For some states, everything depends on the vehicle’s level of automation [18,19].

Market size and trends

Despite challenges in legislation and beyond, the AV industry is lucrative. In the U.S., the current AV market is estimated to be valued at around $4 billion, with projections to become a $186 billion industry by 2030 [2]. It is estimated that by 2035, the AV industry could generate a whopping $300-$400 billion in revenue [26].

While growth is inevitable, it’s difficult to say when cars will be fully autonomous. While researchers, manufacturers, and industry experts differ in their projections, the majority agree that level 5 AVs won’t hit the market until at least 2030, with many projecting 2040 [6, 4]. Even then, we’re only likely to see these vehicles allowed in certain “geofenced” areas, such as corporate and college campuses or theme parks [4]. It’s likely that the use of AVs on major roads will still require human-driver supervision [4].

In looking at the technology available, consumer buy-in, and regulatory support, projections show that at worst, 4% of new passenger cars will be sold with level 3 or higher by 2030 and 17% by 2035 [26]. At best, these numbers increase to 20% by 2030 and 57% by 2035 [26].

Conclusion

While projections may vary and lawmakers debate the appropriate methods of regulation, one thing is clear: the cars of the future are upon us. AVs are already in development and being tested on U.S. roads as we speak. While some may be wary at the thought of relinquishing control on the road, others are excited by the thought of the safety and efficiency that AVs can potentially introduce to driving as we know it. We hope that this article serves to educate you in order to form a more rounded opinion on this emerging technology.

Sources:

Jason L Arthur

Jason Arthur is a data junkie, writer, veteran amateur racecar driver and motorsport photographer. He is the co-founder of LookupaPlate, a collaborative platform to report bad drivers plying on American roads. He is also building a blockchain-based vehicle data marketplace (in stealth mode) and is an adviser to several startups. Jason has been tracking the automotive industry since the 1990s and has a disturbingly deep obsession with the automotive world, and loves to explore whatever roads he can find. From high-speed racing on the circuit to off-road exploration, Jason has an insatiable appetite for adventure. Jason has written for numerous publications, including Autocar Magazine, Motorsport Magazine, and Road & Track.