General Assembly II Measures to Safely Implement and Monitor Autonomous Cars

Size: px

Start display at page:

Download "General Assembly II Measures to Safely Implement and Monitor Autonomous Cars"

Rosamond Walters
5 years ago
Views:

TIANMUN 2018 General Assembly 2 1 FORUM: General Assembly II ISSUE: Measures to Safely Implement and Monitor Autonomous Cars STUDENT OFFICER: Carissa Wu POSITION: President of General Assembly II

1 TIANMUN 2018 General Assembly 2 1 FORUM: General Assembly II ISSUE: Measures to Safely Implement and Monitor Autonomous Cars STUDENT OFFICER: Carissa Wu POSITION: President of General Assembly II Introduction The development of autonomous cars, contrary to common belief, is already incredibly far along. Self-driving vehicles are already on roads in several countries, a couple including the USA, Greece, and Singapore. With autonomous cars, there is massive potential for improvements in car accident rates, transportation accessibility, and time efficiency. Essentially, autonomous cars bring an entirely new mode of transportation, operated by artificial intelligence (AI) instead of humans. As such, there is a great need for countries to consider how to safely implement this technology into their infrastructures, especially during this crucial period of development. As the technology improves rapidly, the urgency for countries to set regulations on autonomous cars also grows. Countries need to quickly address the regulations and systems to combat for problems already surfacing. Some major issues that have arisen in the past couple of years with autonomous cars include concerns with safety (AI hacking, competency of programs, misleading marketing of programs), a lack of legal frameworks, and the safe development of autonomous vehicles. Background The race for a fully autonomous car brings incredible innovation and competition within the global vehicle industry because of the vast benefits and potential as a game-changer in transportation. Possibly the most impactful of these benefits is the change it would bring to road fatalities. Currently, over one million lives are lost each year from car accidents globally, with 90% of these deaths caused by human General Motors' autonomous vehicle, Cruise AV error (Wollaston). With AI, millions of lives could be saved. Additionally, self-driving vehicles could make transportation more accessible to the public. Especially for children, the elderly, and the

2 TIANMUN 2018 General Assembly 2 2 handicapped who cannot drive cars, autonomous cars could bring much greater convenience and mobility. Moreover, without the need for a driver, working adults could spend commute time working or relaxing instead of operating their vehicles, bringing great improvements in life quality and efficiency. Though not possible in the near future, platooning could become a norm when autonomous cars are more integrated into transportation infrastructure. Platooning is the concept in which autonomous vehicles could communicate and travel in platoons in order to lessen traffic and decrease travel time for passengers. Artificial intelligence could also improve beyond the natural sensory and mechanical abilities of humans and push car speeds Model demonstrating the concept of "platooning" well beyond the capability of a human driver, likewise shortening travel time drastically. Currently, there are a number of companies deeply involved in the development of autonomous cars. Amongst the leading programs include those of General Motors, Waymo (owned by Alphabet Inc.), Uber, Tesla, Ford, and Honda. However, while these companies are seriously striving for full-automatization in the near future, most vehicles produced today already have some level of AI control already. Common programs installed in most cars today include driver assistance, lane keeping, and other programs to aid drivers in adjusting the car s speed and orientation. The concept of AI controlling cars is not a foreign concept for the auto industry, and as such, the development of fully autonomous cars is not entirely new. There are five levels of automation according to SAE International (Society of Automotive Engineers) in the J3016 guidelines. Having been adopted by the US Department of Transportations and the UN, these guidelines have become an industry standard, used widely by companies. Currently, cars with up to level 3 automation vehicles are being sold to the public and are on public roads. Problems Raised AI Safety Concerns

TIANMUN 2018 General Assembly 2 3 There are a number of safety concerns with autonomous cars, especially because of a number of accidents caused by program errors that have killed both passengers and

3 TIANMUN 2018 General Assembly 2 3 There are a number of safety concerns with autonomous cars, especially because of a number of accidents caused by program errors that have killed both passengers and a pedestrian. Most famous are cases with Tesla and Uber. Tesla s Enhanced Autopilot, which markets as having full self-driving capability, has caused a number of deaths due to program and sensing errors. One Tesla Model X with autopilot sped up and steered into a concrete barrier at 70.8 mph, another crashed into a stationary police car, while a model S crashed into the back of a stopped firetruck. All accidents resulted in driver fatalities and occurred in the USA (Gibbs). Furthermore, risks extend not only to drivers but also pedestrians around self-driving cars. In 2017, a pedestrian was killed by an Uber self-driving car during a test drive in Arizona (United States). What is most worrisome is how though the Uber program had detected the pedestrian, the program had still decided to override the braking, subsequently running over the pedestrian and killing her. Cases such as these raise concerns with the general competency and trustworthiness of autonomous programs to successfully operate a vehicle without posing a risk to passengers, pedestrians, and neighboring cars alike. They additionally highlight the importance of the regulations regarding the safety testing of autonomous programs. The creation and enforcement of regulations to ensure the safety of all parties involved in the development of autonomous cars need to be of paramount importance in order to prevent more deaths. Misleading Marketing of Autonomous Vehicles Currently, there are several car models with partial self-driving capability for sale in the public market. These include Tesla s Enhanced Autopilot and Nissan s ProPilot. However, as can be seen with these before mentioned features, companies tend to market their autonomous programs to give consumers the impression that they can completely trust the car to operate safely. Terms such as Autopilot, ProPilot, Misleading advertising of Tesla's Autopilot full self-driving capability, and being capable of driving autonomously used to advertise these products lull drivers into a sense of false security, encouraging consumers to fully trust the programs to operate the car, when in actuality, these companies have regulations that require the driver to be focused at all times and ready to take over in the case the program fails (Gibbs). This misunderstanding led to the tragic death of a Tesla enthusiast in 2016, when his car registered a white truck in front as part of the sky, thus failing to register its presence and crashing into it at 74

4 TIANMUN 2018 General Assembly 2 4 mph, effectively killing the driver. Theoretically, the driver should have been diligent and prevented the accident when he realized his car was speeding up and heading directly into a truck, but because he felt Tesla s Enhanced Autopilot completely capable of transporting him safely across the highway, failed to maintain the level of alertness required by the company to stay unharmed (Corfield). Lack of Legal Frameworks for Autonomous Cars Because cars capable of full autonomy in driving have only been developed in recent years, many countries lack the proper legal frameworks for this new technology. Below is an informational chart on the current state of the legality of autonomous vehicles in the world as of 2018 ( Global Survey of Autonomous Vehicle Regulations ). State of legislation on autonomous cars around the world As seen above, many countries have legal barriers against cars on public roads without drivers. Reasonably, because of the current developmental status of self-driving cars, governments are hesitant to integrate fully automated cars, to any extent, onto their roads. However, if a government supports the development of these vehicles, there needs to be definite law in place to deal with any potential accidents, namely who to charge blame in the case of an accident the company of the faulty program, the inattentive driver of the car, or neighboring cars. With each situation, there are different conditions, so it is incredibly important that laws address such circumstances to avoid unnecessary confusion.

TIANMUN 2018 General Assembly 2 5 Safe Development of Autonomous Cars One of the key reasons behind many of the accidents is that drivers may not have been attentive to the car s course as required

5 TIANMUN 2018 General Assembly 2 5 Safe Development of Autonomous Cars One of the key reasons behind many of the accidents is that drivers may not have been attentive to the car s course as required by the companies selling the vehicles. Disregarding mislead customers, there have been accidents when test drivers, who should be completely aware of the Uber's self-driving car program installed on a Volvo XC90 SUV, the same model as the vehicle supervised by Rafaela Vasquez programs capabilities, fail to properly take control of the car before it crashes. Often, this is due to the lack of attention paid by the test driver on the road. Perhaps the most famous of these cases was when Rafaela Vasquez, a test driver for Uber, crashed into Elain Herzberg, a homeless woman, who had been crossing the road at night with her bicycle. Footage revealed after the collision showed that for twenty minutes before the crash, Vasquez had been watching The Voice on her phone instead of remotely paying attention to the vehicle s course. It was only seconds after the collision that Vasquez realized what had occurred and hit the brakes manually ( Driver was streaming The Voice when Uber self-driving car crashed, say police. ). As test drivers are paid and contracted to prevent incidents such as these, it is imperative that there be more measures to prevent this from happening and ensure the development of autonomous cars is safe for the public. Preventing the Potential Hacking of AI Programs As is a weakness for any type of artificial intelligence, there is always a possibility if not now, then in the future that AI software can be hacked and controlled externally for malicious purposes. This could risk both the lives of passengers within and outside of the car. Remote control of vehicles could open many doors for criminals to commit theft, harassment, or even murder. This very real danger has been addressed by autonomous car developers, such as General Motors, which has stated they work with many third parties to maintain and advance [their] cybersecurity guidelines and standards ( 2018 Self-Driving Safety Report ). There are many ways to achieve an adequate level of safety for passengers, all of which will most definitely require collaboration to eliminate the risk of programs being hacked.

TIANMUN 2018 General Assembly 2 6 International Actions Creation of the 1968 Vienna Convention on Road Traffic The 1968 Convention on Road Traffic was a treaty ratified by 74-member nations during

6 TIANMUN 2018 General Assembly 2 6 International Actions Creation of the 1968 Vienna Convention on Road Traffic The 1968 Convention on Road Traffic was a treaty ratified by 74-member nations during the Economic and Social Council s Conference on Road Traffic in Notable member nations who have not signed onto the treaty include the USA, China, and the UK (Swinhoe). Of special importance in this treaty is article 8, which had formerly stated that Every driver shall at all times be able to control his vehicle or to guide his animals. Because the driver was required to be in control of the car at all times, the testing or existence of fully autonomous cars at level 3 or higher was not permitted at all in signed countries. However, this article was revised in 2014 to allow autonomous cars on the condition the driver is ready to take over in case of program error ( UNECE paves the way for automated driving by updating UN international convention ). Creation of the Six Levels of Autonomy The six levels of autonomation, defined below in Key Terms, was released by the Society of Automotive Engineers (SAE) International in 2014 in response to the advancing development of autonomous vehicles. SAE International s definition for the levels of autonomy have been adopted as an industry standard, and even some governments (such as the USA). Key Players United States of America The United States is a huge center for testing of autonomous cars in the world. Companies such as Waymo, General Motors, Tesla, and Uber test their cars in several states. As such, the USA should be actively involved in discussions on this issue, especially because it has suffered the most deaths from autonomous vehicles as well. The USA is very open to autonomous vehicles, shown below by the legal status of autonomous cars (at least on some level) in 24 of the 50 states. Legal status of autonomous cars in the USA by state

TIANMUN 2018 General Assembly 2 7 Netherlands The Netherlands has invested a considerable sum on transforming their transportation infrastructure to support electric vehicles Though the Dutch do not

7 TIANMUN 2018 General Assembly 2 7 Netherlands The Netherlands has invested a considerable sum on transforming their transportation infrastructure to support electric vehicles Though the Dutch do not necessarily hold the same views, the government of the Netherlands has invested heavily in development for autonomous trucks and has pushed EU legislation to increase progress on self-driving vehicles. Furthermore, the Netherlands is arguably the country with the most well-prepared infrastructure to accommodate for autonomous vehicles, mainly because of the many charging stations and other infrastructure made for the many electric vehicles (a popular mode of transportation in the Netherlands and subsidized heavily by the Dutch government). Likewise, the Netherlands should take an active part in this discussion. Possible Solutions Gathering Insight to Aid the Creation of Regulations to Integrate Autonomous Cars As autonomous cars become increasingly viable as a mode of transportation for the general public, countries will be pressured to create regulations and laws to ensure the safe implementation of autonomous cars onto public roads. A collaboration between member-nations to share wisdom could be of great help, especially to countries with no prior experience handling autonomous cars or potential issues that could arise from them. Facilitating Collaboration Between Developers and Member-Nations to Prevent Collateral Damage During Program Development Collaboration between involved parties to share information on errors and problems during test drives, at least to some extent, could be of huge help to the generally forward direction of the development of autonomous vehicles. Program security information combatting potential hackers could also be shared to combine and strengthen efforts against such dangers. The sharing of data could potentially prevent the same error from occurring in vehicles from different firms. Instead, this

8 TIANMUN 2018 General Assembly 2 8 information could be corroborated for open viewing so all developers can further the safety of their cars. Government Support on the Development of Autonomous Cars For countries valuing the potential benefits of autonomous cars and wishing to support the creation of the technology, subsidies, tax cuts, and government grants could greatly quicken the pace of development. This kind of government action can provide incentives and much-needed resources to enable developers to create safe autonomous car programs at a faster rate. Member-nations could also potentially come together to create an international effort if so inclined in order to accelerate the rate of development. Glossary Six Levels of Automation Level 0: No Automation Level 1: Driver Assistance Level 2: Partial Automation Level 3: Conditional Automation Level 4 High Automation Level 5: Full Automation Self-driving Vehicle Visual explanation of the six levels of automation An automotive vehicle capable of driving without the full attention of the driver. However, the term is loosely used to define any self-driving car, including those that may still require the attention of the driver in the case of program error. Platooning The concept in which the AI programs of autonomous vehicles could potentially, through wireless networks, communicate and coordinate with one another to travel in platoons in order to lessen traffic and decrease travel time for passengers. Autonomous car A vehicle that is capable of fully operating a car with artificial intelligence. Supervision by a driver is permitted for the sake of this topic.

9 TIANMUN 2018 General Assembly 2 9 Sources Bliss, Laura. You re Thinking About Autonomous Vehicles Wrong. Citylab, 8 June 2018, Driver was streaming The Voice when Uber self-driving car crashed, say police. The Guardian, 22 June 2018, Gibbs, Samuel. Motorists are being misled by autonomous driving aids report The Guardian, 11 June 2018, aids-tesla-nissan-report. Global Survey of Autonomous Vehicle Regulations. Synced, 15 March 2018, syncedreview.com/2018/03/15/global-survey-of-autonomous-vehicle-regulations/. Corfield, Gareth. Tesla death smash probe: Neither driver nor autopilot saw the truck. The Register, 20 June 2017, Halsey, Ashley, III. Most people expect driverless cars to become common, and they worry about it. Washington Post, 4 June 2018, people-expect-driverless-cars-tobecome-common-and-they-worry-about-it/2018/06/03/fa213ea0-64ed-11e8-a768- ed043e33f1dc_story.html?utm_term=.c06f582ccb06. Accessed 26 June Hyatt, Kyle. Paukert, Chris. Self-driving cars: A level-by-level explainer of autonomous vehicles. Road Show, CBS Interactive Inc., 29 March 2018, Litman, Todd. "Autonomous Vehicle Implementation Predictions." Victoria Transport Policy Institute, 26 June 2018, Accessed 27 June Small, Andrew. Self-Driving Cars Still Have a Lot to Learn. Citylab, 16 April 2018, Swinhoe, Dan. UN legalizes Autonomous vehicles. IDG Connect, 24 March 2016, UNECE paves the way for automated driving by updating UN international convention. United Nations Economic Commission for Europe, 23 March 2016,

10 TIANMUN 2018 General Assembly automated-driving-by-updating-un-international-convention/doc.html. Walker, Jon. The Self-Driving Car Timeline Predictions from the Top 11 Global Automakers. Tech Emergence, 29 May 2018, Weekly Brief: Self-driving cars score big win with UN. TU-Automotive, 26 May 2014, analysis.tuauto.com/other/weekly-brief-self-driving-cars-score-big-win-un. Woollaston, Victoria. "How do Google s driverless cars work?" alphr, 4 April 2016, Accessed 25 June "2018 Self-Driving Safety Report" General Motors, 2018,

Uber autonomous vehicle death raises questions for UK law review

Uber autonomous vehicle death raises questions for UK law review Alex Scroxton Networking Editor 22 Mar 2018 9:30 Uber suspended testing of autonomous vehicles in North America after one of its cars struck