Self-driving car

Features such as keeping the car within its lane, speed controls, and emergency braking are termed driver assistance, because while they handle some driving tasks, they require a human driver.

Organizations such as AAA provide standardized naming conventions for features such as automated lane keeping support (ALKS). The Association of British Insurers stated that the usage of the word autonomous in marketing to be dangerous because car ads make motorists think "autonomous" and "autopilot" imply that the driver can rely on the car to control itself, even though they rely on the driver to ensure safety.

Despite offering something called Full Self-Driving, Tesla stated that its offering is not completely autonomous.[54] In the United Kingdom, a fully self-driving car is defined as a car registered in a specific list, rather than a set of features.[55] Proposals to adopt aviation automation terminology for cars have not prevailed.[56]

According to SMMT, "There are two clear states – a vehicle is either assisted with a driver being supported by technology or automated where the technology is effectively and safely replacing the driver."[57]

Many projects have automated (made automatic) some aspect of driving. Some required aids in the environment, such as magnetic strips in roadways. Autonomous control implies performance under environmental uncertainty, along with the ability to compensate for errors without external intervention.[58]

One approach is to pool information across multiple vehicles. This can be done locally, to e.g., form a convoy or more widely, e.g., to traffic-optimize a route.

Euro NCAP defined autonomous as "the system acts independently of the driver to avoid or mitigate the accident", which implies the autonomous system is not the driver.[59]

In Europe, the words automated and autonomous might be used together. For instance, Regulation (EU) 2019/2144 supplied:[60]

• "automated vehicle" means a motor vehicle designed and constructed to move autonomously for certain periods of time without continuous driver supervision but in respect of which driver intervention is still expected or required;[60]
• "fully automated vehicle" means a motor vehicle that has been designed and constructed to move autonomously without driver supervision;[60]

In British English, the word automated alone might have several meanings, such as in the sentence: "Thatcham also found that the automated lane keeping systems could only meet two out of the twelve principles required to guarantee safety, going on to say they cannot, therefore, be classed as 'automated driving', instead it claims the tech should be classed as "assisted driving".":[61] The first occurrence of the "automated" word refers to an Unece automated system, while the second refers to the British legal definition of an automated vehicle. British law interprets the meaning of "automated vehicle" based on the interpretation section related to a vehicle "driving itself" and an insured vehicle.[62]

To enable a car to travel without a driver within the vehicle, some companies use a remote driver.

According to SAE J3016,

Some driving automation systems may indeed be autonomous if they perform all of their functions independently and self-sufficiently, but if they depend on communication and/or cooperation with outside entities, they should be considered cooperative rather than autonomous.

PC Magazine defined a self-driving car as "a computer-controlled car that drives itself".[63] The Union of Concerned Scientists used "cars or trucks in which human drivers are never required to take control to safely operate the vehicle. Also known as autonomous or 'driverless' cars, they combine sensors and software to control, navigate, and drive the vehicle."[64]

The British Automated and Electric Vehicles Act 2018 law defines a vehicle as "driving itself" if the vehicle "is operating in a mode in which it is not being controlled, and does not need to be monitored, by an individual".[65]

Another British definition adopts "Self-driving vehicles are vehicles that can safely and lawfully drive themselves."[66]

Tesla Autopilot is classified as an SAE Level 2 system.[67][68]

A classification system with six levels – ranging from fully manual to fully automated systems – was published in 2014 by SAE International as J3016, Taxonomy and Definitions for Terms Related to On-Road Motor Vehicle Automated Driving Systems; the details are revised periodically.[13] This classification is based on the role of the driver, rather than the vehicle's capabilities, although these are loosely related. In the United States in 2013, the National Highway Traffic Safety Administration (NHTSA) released its original formal classification system. After SAE updated its classification in 2016, called J3016_201609,[69] NHTSA adopted the SAE standard,[70][71][72][73]

"Driving mode" is used as "a type of driving scenario with characteristic dynamic driving task requirements (e.g., expressway merging, high speed cruising, low speed traffic jam, closed-campus operations, etc.)"[1][74]

• Level 0: The automated system issues warnings and may momentarily intervene, but has no sustained vehicle control.
• Level 1 ("hands on"): The driver and the vehicle share control. Examples are systems where the driver controls steering and the automated system controls engine power to maintain a set speed (cruise control) or engine and brake power to maintain and vary speed (adaptive cruise control). The driver must always be ready to retake control. Lane keeping (LK) Type II is a further example of Level 1 self-driving. Automatic emergency braking, which alerts the driver to a potential crash and applies the brakes is a Level 1 feature, according to Autopilot Review magazine.[75]
• Level 2 ("hands off"): The automated system takes full control of the vehicle: accelerating, braking, and steering. The driver must monitor the driving and be prepared to intervene immediately at any time. The shorthand "hands off" is not literal – contact between hand and wheel is often mandatory during SAE 2 driving, to confirm that the driver is ready to intervene. The driver's eyes may be monitored to confirm that the driver is attentive. True hands off driving is sometimes unofficially termed level 2.5. A common example is ACC combined with LK, such as "Super-Cruise" in the Cadillac CT6 or the F-150's BlueCruise.[76]
• Level 3 ("eyes off"): The driver can safely turn their attention from driving. The driver must still be prepared to intervene within a time interval specified by the manufacturer, when called upon by the vehicle. This level can be thought of as a co-driver that alerts the driver in an orderly fashion when handing off control. An example would be a Traffic Jam Chauffeur[77] (a car satisfying the Automated Lane Keeping Systems (ALKS) regulations).[78]
• Level 4 ("mind off"): No driver attention is required for safety, allowing the driver to sleep or change seats. However, self-driving is supported only in specific areas (geofenced) or under specific circumstances. Outside of these areas/circumstances, the vehicle must be able to safely abort the trip, e.g. stop and park. An example would be a robotaxi or delivery service that covers specific locations, possibly also time-of-day-limited. Automated valet parking is another example.
• Level 5 ("steering wheel optional"): No human intervention is required under any circumstances, such as long-distance trucking.

The SAE Automation Levels have been criticized for their technological focus. It has been argued that the structure of the levels suggests that automation increases linearly and that more automation is better, which may not always be the case.[79] The SAE Levels also do not account for changes that may be required to infrastructure[80] and road user behavior.[81][82]

Several classifications have been proposed to deal with ADAS technology. One such proposal is to adopt these categories: navigation, path planning, perception, and car control.[83]

Even video games have been used as a platform to test autonomous vehicles.[84]

Navigation involves the use of maps to define a path between origin and destination. Hybrid navigation is the use of multiple navigation systems.

Sensor technologies including combinations of cameras, LiDAR, radar, audio, and ultrasound have been applied to the task of understanding the environment surrounding a vehicle,[85] GPS, and Inertial measurement.[86][87] Some systems use Bayesian simultaneous localization and mapping (SLAM) algorithms. Waymo at one point used SLAM, and added detection and tracking of other moving objects (DATMO), to handle potential obstacles.[88][89] Other systems use roadside real-time locating system (RTLS) technologies to aid localization. Tesla uses eight cameras to create a bird's-eye view of the surroundings and categorize the objects within it.[90]

Maps are necessary for navigation. Map sophistication varies from simple graphs that show which roads connect to each other, with details such as one-way vs two-way, to those that are highly detailed, with information about lanes, traffic controls, roadworks, and more.[85] Researchers at the MITComputer Science and Artificial Intelligence Laboratory (CSAIL) developed a system called MapLite, which allowed self-driving cars to drive without using 3D maps. The system combines the GPS position of the vehicle, a "sparse topological map" such as OpenStreetMap (i.e. having 2D features of the roads only), and a series of sensors that observe road conditions.[91] One issue with highly-detailed maps is keeping them updated as the world changes. Vehicles that can operate with less-detailed maps to some extent require less-frequent updates.

Control systems typically combine data from multiple sensors.[92] Self-driving cars often combine cameras, LiDAR, and radar. Multiple sensors provide a more complete view of the surroundings and can be used to cross-check each other to correct errors.[93]

Path planning finds a sequence of segments that a vehicle can follow from origin to destination. Two techniques used for path planning are graph-based search and variational-based optimization techniques. Graph-based techniques can make harder decisions such as how to pass another vehicle/obstacle. Variational-based optimization techniques require a higher level of planning in setting restrictions on the vehicle's driving corridor to prevent collisions.[94] The large scale path of the vehicle can be determined by using a voronoi diagram, an occupancy grid mapping, or with a driving corridors algorithm. The latter allows the vehicle to locate and drive within open space that is bounded by lanes or barriers.[95]

Drive by wire is the use of electrical or electro-mechanical systems for performing vehicle functions traditionally achieved by mechanical linkages.

Driver monitoring is used to assess the driver's attention and alertness. Techniques in use include eye monitoring, and requiring the driver to maintain torque on the steering wheel.[96]

Vehicles can potentially benefit from communicating with others to share information about traffic, road obstacles, to receive map and software updates, etc.[97][98][85]

ISO/TC 22 specifies in-vehicle transport information and control systems,[99] while ISO/TC 204 specifies information, communication and control systems in surface transport.[100] International standards have been developed for ADAS functions, connectivity, human interaction, in-vehicle systems, management/engineering, dynamic map and positioning, privacy and security.[101]

Software controls the vehicle, and can provide entertainment and other services. Some vehicles can acquire updates to that software over the internet. In March 2021, UNECE regulation on software update and software update management systems was published.[102]

Mobileye's mathematical model, "Responsibility-Sensitive Safety (RSS)",[103] is undergoing standardization as "IEEE P2846: A Formal Model for Safety Considerations in Automated Vehicle Decision Making".[104]

In 2022, a research group of National Institute of Informatics (NII, Japan) expanded RSS and developed "Goal-Aware RSS" to make RSS rules possible to deal with complex scenarios via program logic.[105]

The primary obstacle to ACs is the advanced software and mapping required to make them work safely across the wide variety of conditions that drivers experience.[106] In addition to handling day/night driving in good and bad weather[107] on roads of arbitrary quality, ACs must cope with other vehicles, road obstacles, poor/missing traffic controls, flawed maps, and handle endless edge cases, such as following the instructions of a police officer managing traffic at a crash site.

Other obstacles include cost, liability,[108][109] consumer reluctance,[110] potential ethical dilemmas,[111][112] security,[113][114][115][116] privacy,[107] and legal/regulatory framework.[117] Further, AVs could automate the work of professional drivers, eliminating many jobs, which could slow acceptance.[118]

Tesla calls its Level 2 ADAS "Full Self-Driving (FSD) Beta".[119] US Senators Richard Blumenthal and Edward Markey called on the Federal Trade Commission (FTC) to investigate this marketing in 2021.[120] In December 2021 in Japan, Mercedes-Benz was punished by the Consumer Affairs Agency for misleading product descriptions.[121]

Mercedes-Benz was criticized for a misleading US commercial advertising E-Class models.[122] At that time, Mercedes-Benz rejected the claims and stopped its "self-driving car" ad campaign that had been running.[123][124] In August 2022, the California Department of Motor Vehicles (DMV) accused Tesla of deceptive marketing practices.[125]

In the 2020s, concerns over ACs vulnerability to cyberattacks and data theft emerged.[126]

In 2018 and 2019 former Apple engineers were charged with stealing information related to Apple's self-driving car project.[127][128][129] In 2021 the United States Department of Justice (DOJ) accused Chinese security officials of coordinating a hacking campaign to steal information from government entities, including research related to autonomous vehicles.[130][131] China has prepared "the Provisions on Management of Automotive Data Security (Trial) to protect its own data".[132][133]

Cellular Vehicle-to-Everything technologies are based on 5G wireless networks.[134] As of November 2022, the US Congress was considering the possibility that imported Chinese AC technology could facilitate espionage.[135]

While predicting the behavior of ACs that do not use traditional communications such as hand signals, is a major challenge for human drivers,[136] the real-time prediction of the behavior of other vehicles, pedestrians etc, some of which may be stationary when first noted, is even greater challenge for self-driving cars.[10] Raster-based methods have been replaced by vector-based methods in order to overcome the former's lossy rendering, limited receptive field, and prohibitively high cost. The remaining problem is high level of uncertainty that emerges in trajectory predictions as the prediction timeframe is extended. Also, if data re-normalization and re-encoding are used to update future trajectories each time a self-driving car changes its position, its action is often delayed by 8 milliseconds, potentially causing an accident. Several powerful trajectory prediction models have recently adopted Transformers with factorized attention as their encoders, but their scalability is still limited by the computational complexity of factorized attention. Most recently proposed QCNet model uses a query-centric instead of agent-centric modeling, taking advantage of both anchor-based and anchor-free solutions, with an anchor-free module generating adaptive anchors in a data-driven manner and an anchor-based module refining these anchors based on the scene context. The model injects the relative spatialtemporal positions into the key and value (both Transformer elements) when performing attention-based scene-context fusion.[137]

For ACs that have not achieved L5, the ADAS has to be able to safely accept control from and return it to the driver.[138]

The second challenge is known as risk compensation: as a system is perceived to be safer, on average people engage in riskier behavior. (People who wear seat belts drive faster). ACs suffer from this problem: for example Tesla Autopilot users in some cases stop monitoring the vehicle while it is in control.

In order for people to buy self-driving cars and vote for the government to allow them on roads, the technology must be trusted as safe.[139][140] Automatic elevators were invented in 1900, but did not become common until operator strikes and trust was built with advertising and features such as an emergency stop button.[141][142]

Standards for liability have yet to be adopted to address crashes and other incidents. Does liability rest with the manufacturer or the driver/passenger and does it vary with, e.g., automation level or merely the specific circumstances?[143]

The trolley problem is a thought experiment in ethics. Adapted for ACs, consider an AC carrying a passenger when suddenly a pedestrian steps in its way and the car has to choose between killing the pedestrian or swerving into a wall, killing the passenger.[144] Ethical researchers have suggested deontology (formal rules) and utilitarianism (harm reduction) as applicable.[10][145][146]

Public opinion has been reported to support harm reduction, except that they want the vehicle to prefer them when they are riding in it. However, utilitarian regulations are unpopular.[147]

Privacy-related issues arise mainly from the fact that ACs are connected to the internet. Any connected device offers the potential to be penetrated. This information includes destinations, routes, cabin recordings, media preferences, behavioral patterns, and others.[148][149][150]

Whether existing road infrastructure can support higher levels of automation has not been finalized. The answer may vary across jurisdictions.[151] In March 2023, the Japanese government unveiled a plan to set up a dedicated highway lane for ACs.[152] In April 2023, JR East announced their challenge to raise their self-driving level of Kesennuma Line bus rapid transit (BRT) in rural area from the current Level 2 to Level 4 at 60 km/h.[153]

The testing of vehicles with varying degrees of automation can be carried out either physically, in a closed environment[154] or, where permitted, on public roads (typically requiring a license or permit,[155] or adhering to a specific set of operating principles),[156] or in a virtual environment, i.e. using computer simulations.[157][158] When driven on public roads, automated vehicles require a person to monitor their proper operation and "take over" when needed. For example, New York has strict requirements for the test driver, such that the vehicle can be corrected at all times by a licensed operator; highlighted by Cardian Cube Company's application and discussions with New York State officials and the NYS DMV.[159]

A prototype of Waymo's self-driving car, navigating public streets in Mountain View, California in 2017

In California, self-driving car manufacturers are required to submit annual reports to share how often their vehicles disengaged from autonomous mode during tests.[160] It has been believed that we would learn how reliable the vehicles are becoming based on how often they needed "disengagements".[161]

In 2017, Waymo reported 63 disengagements over 352,545 mi (567,366 km) of testing, an average distance of 5,596 mi (9,006 km) between disengagements, the highest among companies reporting such figures. Waymo also traveled a greater total distance than any of the other companies. Their 2017 rate of 0.18 disengagements per 1,000 mi (1,600 km) was an improvement over the 0.2 disengagements per 1,000 mi (1,600 km) in 2016, and 0.8 in 2015. In March 2017, Uber reported an average of just 0.67 mi (1.08 km) per disengagement. In the final three months of 2017, Cruise (now owned by GM) averaged 5,224 mi (8,407 km) per disengagement over a total distance of 62,689 mi (100,888 km).[162] In July 2018, the first electric driverless racing car, "Robocar", completed a 1.8-kilometer track, using its navigation system and artificial intelligence.[163]

Disengagements
As of 2022, "disengagements" are at the center of the controversy. The problem is that reporting companies have varying definitions of what qualifies as a disengagement, and that definition can change over time.[166][161] Executives of self-driving car companies have criticized disengagements as a deceptive metric, because it does not take into account the higher degree of difficulty navigating urban streets compared with interstates highway.[167]

Compliance
In April 2021, WP.29 GRVA issued the master document on "Test Method for Automated Driving (NATM)".[168]

In October 2021, the Europe's comprehensive pilot test of automated driving on public roads, L3Pilot, demonstrated automated systems for cars in Hamburg, Germany, in conjunction with ITS World Congress 2021. SAE Level 3 and 4 functions were tested on ordinary roads.[169][170] At the end of February 2022, the final results of the L3Pilot project were published.[171]

In November 2022, an International Standard ISO 34502 on "Scenario based safety evaluation framework" was published.[172][173]

Collision avoidance
In April 2022, collision avoidance testing was demonstrated by Nissan.[174][175] Also, Waymo published a document about collision avoidance testing in December 2022.[176]

Simulation and validation
In September 2022, Biprogy released a software system of "Driving Intelligence Validation Platform (DIVP)" as the achievement of Japanese national project "SIP-adus" led by Cabinet Office with the same name of its subproject which is interoperable with Open Simulation Interface (OSI) of ASAM.[177][178][179]

Topics
In November 2021, the California Department of Motor Vehicles (DMV) notified Pony.ai that it was suspending its driverless testing permit following a reported collision in Fremont on 28 October. This incident stands out because the vehicle was in autonomous mode and didn't involve any other vehicle.[180] In May 2022, DMV revoked Pony.ai's permit for failing to monitor the driving records of the safety drivers on its testing permit.[181]

In April 2022, it is reported that Cruise's testing vehicle blocked fire engine on emergency call, and sparked questions about an autonomous vehicle's ability to handle unexpected roadway issues.[182][183]

In November 2022, Toyota gave a demonstration of one of its GR Yaris test car equipped with AI, which had been trained on the skills and knowledge of professional rally drivers to enhance the safety of self-driving cars.[184] Toyota has been using the learnings from the collaborative activities with Microsoft in FIA World Rally Championship since 2017 season.[185]

Pedestrian reaction
In 2023 David R. Large, senior research fellow with the Human Factors Research Group at the University of Nottingham, disguised himself as a car seat in a study to test people's reactions to driverless cars. He said, "We wanted to explore how pedestrians would interact with a driverless car and developed this unique methodology to explore their reactions." The study found that, in the absence of someone in the driving seat, pedestrians trust certain visual prompts more than others when deciding whether to cross the road.[186]

As of November 2021, Tesla's advanced driver-assistance system (ADAS) Autopilot is classified as a Level 2.[187]

On 20 January 2016, the first of five known fatal crashes of a Tesla with Autopilot occurred in China's Hubei province.[188] According to China's 163.com news channel, this marked "China's first accidental death due to Tesla's automatic driving (system)". Initially, Tesla pointed out that the vehicle was so badly damaged from the impact that their recorder was not able to conclusively prove that the car had been on autopilot at the time; however, 163.com pointed out that other factors, such as the car's absolute failure to take any evasive actions prior to the high speed crash, and the driver's otherwise good driving record, seemed to indicate a strong likelihood that the car was on autopilot at the time. A similar fatal crash occurred four months later in Florida.[189][190] In 2018, in a subsequent civil suit between the father of the driver killed and Tesla, Tesla did not deny that the car had been on autopilot at the time of the accident, and sent evidence to the victim's father documenting that fact.[191]

The second known fatal accident involving a vehicle being driven by itself took place in Williston, Florida on 7 May 2016 while a Tesla Model S electric car was engaged in Autopilot mode. The occupant was killed in a crash with an 18-wheel tractor-trailer. On 28 June 2016 the US National Highway Traffic Safety Administration (NHTSA) opened a formal investigation into the accident working with the Florida Highway Patrol. According to NHTSA, preliminary reports indicate the crash occurred when the tractor-trailer made a left turn in front of the Tesla at an intersection on a non-controlled access highway, and the car failed to apply the brakes. The car continued to travel after passing under the truck's trailer.[192][193] NHTSA's preliminary evaluation was opened to examine the design and performance of any automated driving systems in use at the time of the crash, which involved a population of an estimated 25,000 Model S cars.[194] On 8 July 2016, NHTSA requested Tesla Motors provide the agency detailed information about the design, operation and testing of its Autopilot technology. The agency also requested details of all design changes and updates to Autopilot since its introduction, and Tesla's planned updates schedule for the next four months.[195]

According to Tesla, "neither Autopilot nor the driver noticed the white side of the tractor-trailer against a brightly lit sky, so the brake was not applied." The car attempted to drive full speed under the trailer, "with the bottom of the trailer impacting the windshield of the Model S". Tesla also claimed that this was Tesla's first known autopilot death in over 130 million miles (210 million kilometers) driven by its customers with Autopilot engaged, however by this statement, Tesla was apparently refusing to acknowledge claims that the January 2016 fatality in Hubei China had also been the result of an autopilot system error. According to Tesla there is a fatality every 94 million miles (151 million kilometers) among all type of vehicles in the US.[192][193][196] However, this number also includes fatalities of the crashes, for instance, of motorcycle drivers with pedestrians.[197][198]

In July 2016, the US National Transportation Safety Board (NTSB) opened a formal investigation into the fatal accident while the Autopilot was engaged. The NTSB is an investigative body that has the power to make only policy recommendations. An agency spokesman said "It's worth taking a look and seeing what we can learn from that event, so that as that automation is more widely introduced we can do it in the safest way possible."[199] In January 2017, the NTSB released the report that concluded Tesla was not at fault; the investigation revealed that for Tesla cars, the crash rate dropped by 40 percent after Autopilot was installed.[200]

In 2021, NTSB Chair called on Tesla to change the design of its Autopilot to ensure it cannot be misused by drivers, according to a letter sent to the company's CEO.[187]

Google's in-house automated car

Waymo originated as a self-driving car project within Google. In August 2012, Google announced that their vehicles had completed over 300,000 automated-driving miles (500,000 km) accident-free, typically involving about a dozen cars on the road at any given time, and that they were starting to test with single drivers instead of in pairs.[201] In late-May 2014, Google revealed a new prototype that had no steering wheel, gas pedal, or brake pedal, and was fully automated.[202] As of March 2016, Google had test-driven their fleet in automated mode a total of 1,500,000 mi (2,400,000 km).[203] In December 2016, Google Corporation announced that its technology would be spun off to a new company called Waymo, with both Google and Waymo becoming subsidiaries of a new parent company called Alphabet.[204][205]

According to Google's accident reports as of early 2016, their test cars had been involved in 14 collisions, of which other drivers were at fault 13 times, although in 2016 the car's software caused a crash.[206]

In June 2015, Brin confirmed that 12 vehicles had suffered collisions as of that date. Eight involved rear-end collisions at a stop sign or traffic light, two in which the vehicle was side-swiped by another driver, one in which another driver rolled through a stop sign, and one where a Google employee was controlling the car manually.[207] In July 2015, three Google employees suffered minor injuries when their vehicle was rear-ended by a car whose driver failed to brake at a traffic light. This was the first time that a collision resulted in injuries.[208] On 14 February 2016 a Google vehicle attempted to avoid sandbags blocking its path. During the maneuver it struck a bus. Google stated, "In this case, we clearly bear some responsibility, because if our car hadn't moved, there wouldn't have been a collision."[209][210] Google characterized the crash as a misunderstanding and a learning experience. No injuries were reported in the crash.[206]

In March 2018, Elaine Herzberg died after being hit by a self-driving car being tested by Uber's Advanced Technologies Group (ATG) in the US state of Arizona. There was a safety driver in the car. Herzberg was crossing the road about 400 feet from an intersection.[211] This marks the first time an individual is known to have been killed by an autonomous vehicle, and the incident raised questions about regulation of the self-driving car industry.[212] Some experts said a human driver could have avoided the fatal crash.[213] Arizona governor Doug Ducey suspended the company's ability to test and operate its automated cars on public roadways citing an "unquestionable failure" of the expectation that Uber make public safety its top priority.[214] Uber then stopped self-driving tests in California until it was issued a new permit in 2020.[215][216]

In May 2018, the US National Transportation Safety Board (NTSB) issued a preliminary report.[217] The final report 18 months later determined that the immediate cause of the accident was the safety driver's failure to monitor the road because she was distracted by her phone. However, Uber ATG's "inadequate safety culture" contributed to the crash. The report noted from the post-mortem that the victim had "a very high level" of methamphetamine in her body.[218] The board also called on federal regulators to carry out a review before allowing automated test vehicles to operate on public roads.[219][220]

In September 2020, the backup driver, Rafael Vasquez, was charged with negligent homicide, because she did not look at the road for several seconds while her phone was streaming The Voice broadcast by Hulu. She pleaded not guilty and was released to await trial. Uber does not face any criminal charge because in the USA there is no basis for criminal liability for the corporation. The safety driver is assumed to be responsible for the accident, because she was in the driving seat in a capacity to avoid an accident (like in a Level 3). The trial was originally planned for February 2021[221] but is now scheduled to begin in June 2023.[222]

On 9 November 2017, a Navya Arma automated self-driving bus with passengers was involved in a crash with a truck. The truck was found to be at fault of the crash, reversing into the stationary automated bus. The automated bus did not take evasive actions or apply defensive driving techniques such as flashing its headlights, or sounding the horn. As one passenger commented, "The shuttle didn't have the ability to move back. The shuttle just stayed still."[223]

On 12 August 2021, a 31-year-old Chinese man was killed after his NIO ES8 collided with a construction vehicle. NIO's self-driving feature is still in beta and cannot yet deal with static obstacles.[224] Though the vehicle's manual clearly states that the driver must take over when nearing construction sites, the issue is whether the feature was improperly marketed and unsafe. Lawyers of the deceased's family have also called into question NIO's private access to the vehicle, which they argue may lead to the data ending up forged.[225]

On 26 August 2021, a Toyota e-Palette, a mobility vehicle used to support mobility within the Athletes' Village at the Olympic and Paralympic Games Tokyo 2020, collided with a visually impaired pedestrian about to cross a pedestrian crossing.[226] The Toyota bus service was suspended after the accident, and resumed on 31 August 2021 with improved safety measures.[227]

In a 2011 online survey of 2,006 US and UK consumers by Accenture, 49% said they would be comfortable using a "driverless car".[228]

A 2012 survey of 17,400 vehicle owners by J.D. Power and Associates found 37% initially said they would be interested in purchasing a "fully autonomous car". However, that figure dropped to 20% if told the technology would cost US$3,000 more.[229]

In a 2012 survey of about 1,000 German drivers by automotive researcher Puls, 22% of the respondents had a positive attitude towards these cars, 10% were undecided, 44% were skeptical and 24% were hostile.[230]

A 2013 survey of 1,500 consumers across 10 countries by Cisco Systems found 57% "stated they would be likely to ride in a car controlled entirely by technology that does not require a human driver", with Brazil, India and China the most willing to trust automated technology.[231]

In a 2014 US telephone survey by Insurance.com, over three-quarters of licensed drivers said they would at least consider buying a self-driving car, rising to 86% if car insurance were cheaper. 31.7% said they would not continue to drive once an automated car was available instead.[232]

In a February 2015 survey of top auto journalists, 46% predicted that either Tesla or Daimler would be the first to the market with a fully autonomous vehicle, while (at 38%) Daimler was predicted to be the most functional, safe, and in-demand autonomous vehicle.[233] In 2015, a questionnaire survey by Delft University of Technology explored the opinion of 5,000 people from 109 countries on automated driving. Results showed that respondents, on average, found manual driving the most enjoyable mode of driving. 22% of the respondents did not want to spend any money for a fully automated driving system. Respondents were found to be most concerned about software hacking/misuse, and were also concerned about legal issues and safety. Finally, respondents from more developed countries (in terms of lower accident statistics, higher education, and higher income) were less comfortable with their vehicle transmitting data.[234] The survey also gave results on potential consumer opinion on interest of purchasing an automated car, stating that 37% of surveyed current owners were either "definitely" or "probably" interested in purchasing an automated car.[234]

In 2016, a survey in Germany examined the opinion of 1,603 people, who were representative in terms of age, gender, and education for the German population, towards partially, highly, and fully automated cars. Results showed that men and women differ in their willingness to use them. Men felt less anxiety and more joy towards automated cars, whereas women showed the exact opposite. The gender difference towards anxiety was especially pronounced between young men and women but decreased with participants' age.[235]

In 2016, a PwC survey, in the United States, showing the opinion of 1,584 people, highlights that "66 percent of respondents said they think autonomous cars are probably smarter than the average human driver". People are still worried about safety and mostly the fact of having the car hacked. Nevertheless, only 13% of the interviewees see no advantages in this new kind of cars.[236]

In 2017, Pew Research Center surveyed 4,135 US adults from 1–15 May and found that many Americans anticipate significant impacts from various automation technologies in the course of their lifetimes—from the widespread adoption of automated vehicles to the replacement of entire job categories with robot workers.[237]

In 2019, results from two opinion surveys of 54 and 187 US adults respectively were published. A new standardized questionnaire, the autonomous vehicle acceptance model (AVAM) was developed, including additional description to help respondents better understand the implications of different automation levels. Results showed that users were less accepting of high autonomy levels and displayed significantly lower intention to use highly autonomous vehicles. Additionally, partial autonomy (regardless of level) was perceived as requiring uniformly higher driver engagement (usage of hands, feet and eyes) than full autonomy.[238]

In 2022, research by safety charity Lloyd's Register Foundation uncovered that only a quarter (27%) of the world's population would feel safe in self-driving cars.[239]

SAE Level 2 features are available as part of the advanced driver-assistance system (ADAS) abilities in many commercially available vehicles. These systems often require a subscription to an ongoing service or paid upgrade with the car purchase.

Ford started offering the "BlueCruise" service on certain electric and gas-powered vehicles in 2022; it is named "ActiveGlide" in Lincoln vehicles. The system provides features such as lane centering, street sign recognition and hands-free highway driving on more than 130,000 miles of divided highways in the US. The version 1.2 update of the service was released in September 2022, and added features like hands-free lane changing, in-lane repositioning, and predictive speed assist.[243][244] In April 2023 BlueCruise technology was approved in the UK, for use on certain motorways. The technology will at first only be available for 2023 models of Ford's electric Mustang Mach-E SUV.[245]

Tesla vehicles are equipped with hardware that Tesla claims will allow full self-driving in the future. The Tesla Autopilot suite of ADAS features are included in all Tesla vehicle models. More advanced driving features are available at an extra cost, under the "Enhanced Autopilot" and "Full Self-Driving" names. The marketing names have been criticized as misleading, as all Tesla ADAS features provide only Level 2 capabilities.[246]

General Motors is developing the "Ultra Cruise" ADAS system, that will be a dramatic improvement over their current "Super Cruise" system. Ultra Cruise will cover "95 percent" of driving scenarios on 2 million miles of roads in the US, according to the company. The system hardware in and around the car includes multiple cameras, short- and long-range radar, and a LiDAR sensor, and will be powered by the Qualcomm Snapdragon Ride Platform. The luxury Cadillac Celestiq electric vehicle will be one of the first vehicles to feature Ultra Cruise.[247]

In 2017, BMW had been trying to make 7 Series as an automated car in public urban motorways of the United States, Germany and Israel before commercializing them in 2021.[248] Although it was not realized, BMW is still preparing 7 Series to become the next manufacturer to reach Level 3 in the second half of 2022.[249][250]

In September 2021, Stellantis has presented its findings from a pilot programme testing Level 3 autonomous vehicles on public Italian highways. Stellantis's Highway Chauffeur claims Level 3 capabilities, which was tested on the Maserati Ghibli and Fiat 500X prototypes.[251] Stellantis is going to roll out Level 3 capability within its cars in 2024.[252]

In January 2022, Polestar, a Volvo Cars' brand, indicated its plan to offer Level 3 autonomous driving system in the Polestar 3 SUV, Volvo XC90 successor, with technologies from Luminar Technologies, Nvidia, and Zenseact.[253]

In the same month, Bosch and the Volkswagen Group subsidiary CARIAD released a collaboration for autonomous driving up to level 3. This Joint development targets to be explored and evalauted for Level 4.[254]

As of February 2022, Hyundai Motor Company is in the stage of enhancing cybersecurity of connected cars to put Level 3 self-driving Genesis G90 on Korean roads.[255]

In December 2022, Honda announced that it will enhance its Level 3 technology to function at any speed below legal limits on highways by 2029.[256][257]

In early 2023, Mercedes-Benz received authorization for its Level 3 Drive Pilot in Nevada,[258] and plans to apply for approval in California by mid-2023.[259] Drive Pilot is planned to be available in the US market as an option for some models in the second half of 2023.

Cruise and Waymo offer limited robotaxi services in a handful of American cities, as fully autonomous vehicles without any human safety drivers in the vehicles.[260]

On 1 April 2023 in Japan, Level 4 legal scheme of the amended "Road Traffic Act" was nation-wide enforced, and one service level-upped to the Level 4 service.[261] The approved self-driving shuttle is "ZEN drive Pilot Level 4" custom-made by AIST.[262]

In July 2020, Toyota started testing with public demonstration rides on Lexus LS (fifth generation) based TRI-P4 with Level 4 capability.[263] In August 2021, Toyota operated potentially Level 4 service using e-Palette around the Tokyo 2020 Olympic Village.[264]

In September 2020, Mercedes-Benz introduced world's first commercial Level 4 Automated Valet Parking (AVP) system named Intelligent Park Pilot for its new S-Class. The system can be pre-installed but is conditional on future national legal approval.[265][266]

In September 2021, Honda started testing programme toward launch of Level 4 mobility service business in Japan under collaboration with Cruise and General Motors, using Cruise AV.[267] In October 2021 at World Congress on Intelligent Transport Systems, Honda presented that they are already testing Level 4 technology on modified Legend Hybrid EX.[268] At the end of the month, Honda explained that they are conducting verification project on Level 4 technology on a test course in Tochigi prefecture. Honda plans to test on public roads in early 2022.[269]

In February 2022, General Motors and Cruise have petitioned NHTSA for permission to build and deploy a self-driving vehicle, the Cruise Origin, which is without human controls like steering wheels or brake pedals. The car was developed with GM and Cruise investor Honda, and its production is expected to begin in late 2022 in Detroit at GM's Factory Zero.[270][271] As of April 2022, the petition is pending.[272]

In April 2022, Honda unveiled its Level 4 mobility service partners to roll out in central Tokyo in the mid-2020s using the Cruise Origin.[273] By September 2022, Japan version prototype of Cruise Origin for Tokyo was completed and started testing.[274]

In January 2023, Holon, the new brand from the Benteler Group, unvield its self-driving shuttle autonomous during the Consumer Electronics Show (CES) 2023 in Las Vegas. The company claims the vehicle is the world's first Level 4 shuttle built to automotive standard. Production of the Holon mover is scheduled to start in the US at the end of 2025.[275]