Public Perceptions of Self-driving Cars: The Case of ...

[Pages:21]1 Public Perceptions of Self-driving Cars:

2 The Case of Berkeley, California

3 4 Daniel Howard 5 University of California, Berkeley ? Department of City and Regional Planning 6 Master of City Planning, MS Transportation Engineering 2014 (expected) 7 228 Wurster Hall, #1850 8 Berkeley, CA 94720 9 djhoward@berkeley.edu 10 11 Danielle Dai (corresponding author) 12 University of California, Berkeley ? Department of City and Regional Planning 13 Master of City Planning, MS Transportation Engineering 2014 (expected) 14 228 Wurster Hall, #1850 15 Berkeley, CA 94720 16 ddai@berkeley.edu 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 Prepared for the 93rd Annual Meeting of the Transportation Research Board 41 Original Submission Date: August 1, 2013 42 43 5,656 words + 5 tables or figures (250 x 5) = 6,906 words 44 45 46

Howard and Dai

2

1 ABSTRACT 2 3 Self-driving vehicles represent a technological leap forward that can offer solutions to current 4 transportation problems and dramatically change how people approach mobility. While self5 driving cars have the potential to improve safety and increase quality of life, many people appear 6 reluctant to adopt the technology, because they are uncomfortable with safety, liabilities, and 7 control. Public attitudes toward self-driving cars are increasingly important as the public shapes 8 the demand for the technology, policies that govern them, and future investments in 9 infrastructure. Moreover the nature of the technology means that the truly transformative benefits 10 are only realized once self-driving cars are adopted en masse. We investigate public attitudes 11 toward self-driving cars using the responses of 107 likely adopters in Berkeley, California as a 12 case study. What do these people find most and least attractive about self-driving cars, and how 13 do they envision the inclusion of the technology? Would they adopt this technology and in what 14 form? Do an individual's demographics, existing travel behavior, and relationship to cars and 15 technology affect his or her opinion about self-driving cars? We find that individuals are most 16 attracted to potential safety benefits, the convenience of not having to find parking, and 17 amenities such as multitasking while en route; conversely, individuals were most concerned with 18 liability, the cost of the technology, and losing control of the vehicle. Men are more likely to be 19 concerned with liability, and less likely to be concerned with control than women. Individuals 20 with higher income are most concerned with liability, and those with lower income appear to be 21 more concerned with safety and control. Single-occupancy vehicle commuters and cyclists were 22 most concerned with giving up control. All groups were concerned with costs. We present this 23 case study to inform those creating this technology how self-driving cars will likely be perceived 24 by the public. 25 26 27 28 29 30 31 32 33 34 35 36 37 38

39 40 41 42 43 44 45 46

Howard and Dai

3

1 INTRODUCTION 2 3 Transportation by automobile in the United States is becoming increasingly unsustainable. 4 Rising carbon emissions, increasing congestion, and high traffic accident rates are a few 5 consequences of auto use. Self-driving vehicles offer an alternative form of individualized 6 transportation that can be adapted to reduce such negative impacts. While self-driving cars have 7 great potential to improve the safety, efficiency, and sustainability of our auto-oriented 8 transportation system, many challenges remain, particularly with public perceptions of safety, 9 liabilities, and control. The ability of self-driving vehicles to effect transformative change 10 depends largely on how successful the vehicles are in attracting drivers from private automobiles. 11 Once a critical mass of self-driving vehicles has been established, network benefits and other 12 economies of scale enable environmental, safety, and travel time improvements. Public attitudes 13 toward self-driving cars become increasingly important as the public shapes the demand and 14 market for the cars, the policies that govern them, and future investments in infrastructure. This 15 study investigates attitudes of likely adopters towards self-driving cars using the responses of 16 107 residents of Berkeley, California as a case study. 17 18 BACKGROUND 19 20 The need for a different form of individual transportation 21 22 Since the mid-twentieth century, automobiles have been the dominant mode of travel both within 23 and between cities. While point-to-point travel on demand confers benefits to the individual, it 24 does so at great costs to society. Automobile travel, particularly single occupancy vehicle (SOV) 25 driving, contributes intensely to traffic congestion, traveler delays, and vehicle pollution. Traffic 26 congestion costs the U.S. economy more than $120 billion, and produces 56 billion pounds of 27 CO2 (1). Automobile accidents are a major concern as well. In 2010, there were approximately 28 35,000 vehicle fatalities in the U.S.; 90% of vehicle crashes can be attributed to driver error (2). 29 Moreover, cars consume valuable resources. On average, cars sit unused almost 22 hours out of 30 every day (3). 31 32 Improvements to the transportation system thus far have been incremental and targeted to 33 specific concerns. For example, converting gasoline-powered vehicles to electric drive helps 34 reduce transportation-related carbon emissions, but does little to reduce congestion or motor 35 vehicle fatalities. Intelligent Transportation Systems (ITS), such as variable message signs, 36 promise safety improvements, but do not explicitly address transportation's contribution to 37 climate change. Transportation Demand Management (TDM) strategies like congestion and 38 parking pricing schemes can go a long way to addressing transportation problems. However, 39 these strategies are best used in combination with improvements in technology. 40 41 Self-driving cars represent a technological leap forward that can offer solutions and dramatically 42 change today's transportation network. A self-driving car (also known as an autonomous car, 43 personal automated vehicle, driverless car, or robotic car) is defined as a motor vehicle capable 44 of automated driving and navigating entirely without direct human input. Self-driving cars sense 45 their surroundings with techniques such as radar, GPS, and computer vision. Advanced control

Howard and Dai

4

1 systems interpret electronic sensor information to identify appropriate navigation paths, as well 2 as obstacles and relevant signage (4, 5). 3 4 Self-driving vehicle technology has origins as early as the 1920s, when Achen Motor Company 5 demonstrated a phantom motor car in Milwaukee (6). Other car companies, as well as 6 electronics companies and universities, experimented with self-driving cars with limited success 7 (7, 8). Advancements in self-driving vehicle technology accelerated in the 2000s with the US 8 government sponsored Defense Advanced Research Projects Agency (DARPA) Grand 9 Challenge in 2004. The DARPA Challenge was the first long distance competition for driverless 10 cars and attracted more than one hundred teams in its first year (9). The winning robot of the 11 second Grand Challenge, led by Sebastian Thrun's team at Stanford, sparked the development of 12 Google's self-driving cars (10). As of 2013, many automotive manufacturers are testing 13 driverless car systems, including Audi, BMW, Ford, General Motors, Mercedes-Benz, Nissan, 14 Toyota (11, 12, 13, 14). Based on their perceptions of the desires of potential customers, many of 15 these systems are not fully autonomous, requiring no driver input, but are rather a form of 16 `autopilot' that can be switched on and off by the driver. 17 18 Opportunities for self-driving cars 19 20 Self-driving vehicles and allied technologies such as connected vehicles, ITS, and electric drive 21 vehicles have the power to change modern transportation to become more sustainable, safer, and 22 convenient compared to today's system. Driverless vehicles can allow people to travel on 23 demand and benefit from the economies of scale that come with being part of a larger 24 transportation network. Self-driving cars can address issues of safety, congestion, fuel 25 efficiency, and equity. 26 27 Improvements in safety could be realized soon after widespread adoption of self-driving 28 vehicles. Self-driving car sensors can follow traffic rules and be more alert and responsive than 29 drivers today. In 2010, there were approximately 35,000 vehicle fatalities in the U.S.; many of 30 these fatalities were caused by distracted driving, drunk driving, and other impairments (3). 31 Research on connected vehicles has shown that vehicle-to-vehicle communication systems 32 potentially address 81% of all police-reported vehicle target crashes annually (15). Self-driving 33 technology includes elements of connected vehicle technology and is likely to gain these safety 34 benefits as well. 35 36 Adoption of self-driving vehicles at the city or regional level is likely to result in reduced 37 congestion across the network once the market penetration of these vehicles passes a certain 38 threshold. As self-driving vehicles have not been deployed on such a scale, empirically 39 determined values for optimum market penetration do not exist. However, one can imagine 40 centralized, demand responsive routing will enable self-driving vehicles to choose a route that 41 minimizes delay for all users in the system and "reserve" a spot in the network. These vehicles 42 would then be able to avoid bottlenecks and congestion prone areas before they begin to slow 43 down traffic. Additionally, forming cars into eight car platoons potentially increases the capacity 44 of freeway lanes about 367% at 45 mph, according to a computer simulation of the effects of 45 platooning (20). These capacity improvements would be achieved without having to add lanes, 46 which uses existing roads more efficiently. Likewise, this will increase the overall throughput in

Howard and Dai

5

1 areas where highway expansion is not possible due to physical characteristics of the site (bridges, 2 hillsides, etc), as well as reduce the amount of land that must be ceded to transportation at the 3 expense of other open space and developments. 4 5 Another opportunity for self-driving cars lies with fuel efficiency and environmental benefits. 6 Platooning vehicles offers immediate benefits to efficiency, regardless of fuel type. The cars 7 behind the lead vehicle reap aerodynamic benefits and reduce their energy consumption by up to 8 25% (18). A 2013 Japanese study showed truck platooning improved fuel economy by 15% for 9 heavy commercial vehicles (19). Since self-driving cars can navigate along a highway with a 10 precision that human drivers cannot, it can enable vehicles to be powered in non-conventional 11 ways that can reduce the energy consumption of the transportation sector. Self-driving vehicles 12 can adopt features of On Line Electric Vehicles (OLEVs), electric vehicles that receive their 13 power from the roadway using inductive power transfers. According to a 2007 IEEE paper, 14 "roadway vehicles driven by human operators cannot meet the tolerance demanded by [the] 15 present [OLEV] system, consequently, system performance is compromised" (17). Self-driving 16 cars enable these benefits and other technologies through their ability to precisely position 17 themselves in the roadway or relative to similarly equipped cars. 18 19 Self-driving cars can also positively address issues of equity. Historically, several groups have 20 been excluded from the flexible, convenient and speedy travel provided by automobiles (21, 22, 21 23). The upfront cost of an automobile is high, preventing economically disadvantaged groups 22 from the benefits of a car. Those who do own cars are likely to spend a disproportionate part of 23 their household expenses on auto ownership. Additionally, people younger than the legal driving 24 age, or those who do not have the physical or mental capabilities to drive a car, are restricted 25 from the benefits of automobile use. With self-driving vehicles, these barriers can be lessened as 26 one does not necessarily have to know or be able to drive to reap the benefits. 27 28 From a social perspective, shared driverless vehicles represent the most affordable way for 29 people to access self-driving technology and its associated benefits. Opening the market for self30 driving cars to those who have been excluded from the automobile market could encourage 31 adoption, lower the price of the technology, and enhance network benefits. From an 32 environmental perspective, shared driverless vehicles are a way to reduce the amount of land 33 consumed for parking, as well as the amount of energy and resources into building vehicles. 34 Finally, shared self-driving cars potentially could reduce VMT since their presence in the market 35 will change the way in which people pay for transportation by translating sunk costs into up36 front costs to consumers, discouraging needless trips, encouraging trip-chaining, and sharing of 37 rides. Shared driverless cars or self-driving taxis have the greatest potential environmental 38 benefits and their adoption should be particularly targeted. 39 40 Challenges to adoption of self-driving cars 41 42 Google's co-founder Sergey Brin says that autonomous vehicles will be available to the public 43 by 2017 (26). While self-driving cars are technologically feasible, significant challenges remain 44 with the legal framework, regulatory changes, cost of technology, and issues of control and trust. 45

Howard and Dai

6

1 As of 2013, California, Nevada, Florida, and the District of Columbia have passed laws 2 authorizing companies to test self-driving cars on private and public roads; nine other states have 3 debated similar bills (27). Yet, the legal and regulatory framework still needs work (4). "If the 4 driver, by design, is no longer in control, what happens if the vehicle crashes? The `driver' could 5 well be an innocent bystander or might at least bear lesser liability than drivers do today" (3). 6 Additionally, self-driving vehicles will likely be connected through V2V (Vehicle to Vehicle) or 7 V2I (Vehicle to Infrastructure) technology (3). Such connectivity will require a large investment 8 in infrastructure that will either have to be provided by the public sector, or made profitable 9 enough that a private entity will be able to provide this service. Regulatory changes must be 10 adopted at the state and local levels to address these issues. The National Highway Traffic 11 Safety Administration has begun efforts, having defined five levels of vehicle automation and 12 announced a policy to address safety of self-driving cars (28). 13 14 Cost of the technology is also a significant barrier to many, and without enough of these vehicles 15 on the roads, the network benefits will not be achieved. "The LIDAR system used in the Google 16 car, for example, costs $70,000" (3). As of 2013, the autonomous driving system costs about 17 $150,000 (27). This is more than the average vehicle purchase and illustrates the need to find 18 technological solutions that the market will bear. Since the success of self-driving vehicles 19 depends on the widespread adoption of this technology as a replacement to the automobile, 20 ascertaining public attitudes towards these vehicles is extremely important since it will affect 21 public support for regulation or expenditures, as well as influence demand for a product 22 introduced into the market. 23 24 Control and trust of the technology is also expected to be a significant issue with the public. The 25 technology is very new and its safety record unproven. Few people feel comfortable using an 26 unproven transportation technology on a regular basis, as one can learn from a study of the 27 history of air travel. Although the Wright Brothers flew at Kitty Hawk in 1903, it was not until 28 the 1920s that the United States had a significant passenger airline network (29). Similar 29 concerns about the safety and reliability of self-driving technology have been expressed (4, 12, 30 13, 14). Moreover, people have expressed concerns about giving up control of the car to a 31 machine, and general beliefs that the car is unsafe. Such concerns about safety and reliability of 32 a nascent technology are natural, and we intend to study their effect on the enthusiasm for self33 driving vehicles. 34 35 Self-driving cars are in development and are not yet fully accessible to the public; as such, 36 limited research about public attitudes toward the technology exists. As of 2013, there have been 37 two significant public opinion surveys conducted by private firms. In 2011, Accenture surveyed 38 2,006 consumers in the US and UK. Nearly half of respondents indicated they would be 39 comfortable with using driverless cars, and the other half was more likely to use the technology 40 if they could take back control if needed (24). In 2012, J.D. Power and Associates conducted a 41 survey of 17,400 vehicle owners; 37% said they would be interested in purchasing a fully 42 autonomous car, but the figure dropped to 20% with the introduction of added costs (25). Since 43 the existing literature on self-driving cars is limited, and the data on public attitudes are 44 proprietary, this study collects and contributes new data to the autonomous vehicle discussion. 45 46

Howard and Dai

7

1 METHODOLOGY

2 3 Public attitudes toward self-driving cars become increasingly important as the public shapes the 4 demand and market for the cars, the policies that govern them, and future investments in 5 infrastructure. Our guiding research questions are:

6

7 What does the public find most attractive and least attractive about self-driving technology?

8 How does the public envision the inclusion of this technology in today's network?

9 Would the public adopt self-driving technology, and in what form?

10 Do an individual's existing travel behavior, relationship to cars, and relationship to

11

technology affect his or her opinion about self-driving cars?

12

13 Data & approach 14 15 To understand public perceptions of self-driving vehicle technology, public opinion data is 16 needed. Limited research currently exists, so we developed and implemented surveys for new 17 data. One of the main obstacles to collecting valid data we encountered was that many people did 18 not have enough information about self-driving cars to offer a fully formed opinion. To help 19 control for differences in background knowledge of the subject, we targeted science museum 20 visitors in Berkeley, California. This group was chosen because they were likely adopters, more 21 willing to listen to an informational video designed to ensure all participates had similar 22 information, and the location was accessible to the authors of this unfunded study. We were then 23 able to focus on the effects that differences in demographics and attitudes towards technology 24 and vehicles have on shaping opinion to these cars. 25 26 We administered the study to visitors at the Lawrence Hall of Science in Berkeley, California in 27 spring 2013. The Lawrence Hall of Science is a public science museum and research center that 28 offers hands-on science exhibits, student workshops, professional development seminars, and 29 other educational opportunities. The research study was advertised to visitors at time of 30 admission, and candy was used as an incentive and reward for participation. The study was 31 administered in a group classroom setting, took approximately 15-20 minutes, and included a 32 survey and 10-minute video. The survey has five sections, and the video was shown after the 33 second section. 34 35 Section A asked questions about an individual's travel behavior and household characteristics. 36 How do you usually commute to work? How do you usually travel for leisure? How many 37 people live in your household? Do you own or lease a vehicle? Do you or anyone in your 38 household expect to purchase a vehicle before 2020? Section B included two questions asking 39 participants to describe their relationship to technology and relationship to cars. For example, do 40 you consider yourself an early adopter, eager to try new technologies, or do you have little or no 41 interest in new technologies? Do you view cars as basic transport, or do you want cars with 42 luxury, style, and energy saving technology? The purpose of these questions was to see whether 43 travel behavior, household characteristics, and one's self-identified relationship to technology 44 and cars might affect one's opinion about self-driving cars. For example, we hypothesized that 45 early adopters of new technologies might be more willing to adopt self-driving cars. 46

Howard and Dai

8

1 After the first two sections, participants were shown a video. As self-driving technology is fairly 2 new and not well understood, a video was critical so participants understand how the technology 3 works. Various videos were considered, and the one selected was the most impartial video 4 found, highlighting both benefits and drawbacks of the technology. The video can be found 5 here: . 6 7 After the video, participants were asked a series of questions about their perceptions of self8 driving cars. In this section, participants were asked to select up to three aspects of self-driving 9 technology they found most and least attractive. Additionally, they were asked two questions 10 about the inclusion of this technology in today's transportation network. Would they prefer if 11 self-driving cars operated in the same lanes as normal traffic, or should self-driving cars be 12 separated from other modes of transportation? Would they be willing to support a public bond 13 measure to build new infrastructure, such as special lanes or traffic signals for self-driving cars? 14 In Section D, participants were asked to consider three scenarios in a future of self-driving cars 15 and their willingness to adopt the technology. A total of 107 survey responses were collected 16 along with voluntary demographic information respondents chose to provide. 17 18 Many measures were taken to ensure a quality research approach. First, the study received 19 Institutional Review Board approval to ensure proper human subjects treatment. We vetted the 20 survey with transportation professionals and members of the UC Berkeley faculty, and pretested 21 the survey before data collection. During administration of the study, we were available to 22 clarify any questions and ensure completion of the surveys. While participants were free to 23 decline to answer any question, we followed up with participants who left questions blank to 24 ensure they purposely left the question blank; this ensured accuracy of our responses. Finally, 25 we ensured impartiality as best we could during administration of the surveys by refraining from 26 discussion until after the respondents had completed the survey. 27 28 We analyzed the data collected in the surveys using a logit model for most questions. The 29 frequency with which respondents indicated they would use a self-driving taxi is analyzed using 30 a log-linear regression. The independent variables are dummy coded compared to a control group 31 with the following attributes: Married, white males with a bachelor's degree and an income 32 between $50,000 and $75,000. The control group had a moderate relationship to technology and 33 valued automobiles with green technology, luxury, and style. We selected this control group 34 because it is strongly represented in the sample and provides a good base with which to compare 35 the responses of those with different attitudes and demographic characteristics. 36 37 Limitations 38 39 The future is difficult to predict. Self-driving technology is not widely adopted and how this 40 technology unfolds is still unknown. Self-driving technology may take an entirely different 41 shape than what is familiar to the public, and more information or future events may change 42 people's attitudes. Stated preferences may change depending on new information and other 43 factors. 44 45 While many efforts were made to ensure impartial data collection, there was still bias with 46 survey distribution, self-selection, and non-response. We intercepted visitors at the Lawrence

 ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download