Go up one level Go home

1 General

CAVs are likely to generate significant changes to transport and mobility. They will generate a number of benefits, as well as bring a number of challenges or problems. The balance between benefits and problems will depend partly on the characteristics of the technology, the policies adopted by administrations and how the vehicles are used. In particular, the choice between owning and using them as MaaS will have a significant impact on this balance. As in other cases, the benefits to individuals will not necessarily align with the maximum benefits to society.

The most frequently mentioned benefit of AVs is a reduction in accidents thanks to removing a significant cause of them: human error. It is also claimed that as AVs could drive closer to each other thanks to faster reaction times, they will also improve traffic flow and road capacities. These two benefits may accrue at relatively low levels of automation, namely 2 and 3. For Levels 4 and 5, there will be three additional benefits to users: the ability to undertake other activities whilst travelling, improved mobility for those currently unable to drive and a mixed-blessing solution to parking problems.

The use of AVs as MaaS will introduce an additional set of benefits and challenges. These are outlined here and discussed in detail in the MaaS section of this site. These main benefits are discussed in turn.

2 Safety

(See also Safety)
It has been estimated that AVs may eliminate up to 90% of accidents as they are mostly caused by human error, either directly or as a contributing factor.This may bring highway traffic accident rates to the level of railways and aviation. In the US this has been estimated to prevent up to US$190 billion in damages and health-costs annually and save thousands of lives. This is a very significant and emotional benefit that is likely to encourage companies and authorities to facilitate the adoption of this technology.

Many of the safety benefits will be achieved with driver assistance technology, for example, SAE Levels 2 and 3 of automation. A good deal will depend on how quickly that level and above become part of the vehicle fleet. Currently, we do not have a full understanding of how intermediate levels of penetration will affect accident rates.

There have been many estimates of the number and costs of accidents saved when AVs are adopted. A 2016 Atkins study estimates that in the UK some 1,700 die from traffic accidents annually and that the total cost to the economy is around £16 billion per year. Atkins states that 90% of those accidents are caused (directly or as a contributing factor) by driver error and are, therefore, mostly preventable with AVs.

An early document from the ENO foundation looks into the number and costs of traffic accidents in the US. It estimates that there are some 5.5 million accidents or crashes a year with some 32,300 fatalities. The total cost of these is estimated at US$ 300 billion per year. Driver error was identified as the main contributing factor of these accidents. In over 90% of the cases and these would be eliminated by AVs when they become almost 100% of the fleet. There may be, of course, some new type of accidents – technical failures.

It is more difficult to estimate the reduction in accidents during the transitional period to 100% of the fleet being AVs. In fact, many of these accident savings will be achieved with Level 3 so the market share of interest will be that of Levels 3 to 5 together. This will be reached well before 100% of Level 5. A reduction in the number of accidents and incidents will have a positive effect on traffic flow.

3 Traffic flow improvement, motorways and urban

In addition to the ability to reduce accidents and incidents, AVs may be able to smooth traffic and reduce shock waves as a result of better reaction times. The safe two seconds headway recommended at present implies a capacity of 1,800 cars per lane per hour on a free-flowing motorway. This headway is a roughly one-second reaction time and one second to allow braking. AVs will shorten the reaction time to a small fraction of a second, thus allowing smaller headways, even if the car in front is not autonomous.
Full connectivity, providing information on traffic conditions ahead, will allow even shorter headways. In the US the Highway Capacity Manual allows about 2,200 passenger cars per hour per lane on freeways. At an average speed of 100 km/hr, this implies about 45 metres per vehicle. Good Adaptive Cruise Control (ACC) should be able to reduce this distance. When combined with the intelligence of Levels 2 and 3, there would be an improvement in the number of vehicles that can safely travel on a motorway. Research on how much of the unused space could be safely occupied by AVs thus increasing road capacity varies widely. At this stage, the most that can be achieved is to run micro-simulation models with different assumptions about the headway the technology will permit and the connectivity of CAVs.

Estimating the performance of AVs and CAVs is not straightforward as it depends not only on sensors (under different light and weather conditions) but also on actuators and other elements of the technology.

Some research appears to be very optimistic: a study by researchers at Columbia University predicts that autonomous cars could increase capacity to about 8,200 cars per hour per lane. The study also estimated that with 100% connected vehicles using V2V communication, capacity could reach 12,000 passenger vehicles per hour travelling safely at 120 km/hr. Other researchers report more modest figures, perhaps 50% capacity improvements with 100% penetration of CAVs.

These wide estimates are illustrated in the following figure.
Picture 3.png
Picture 3.png

(Bradley Kloostra and Matthew J. Roorda (2017) Fully autonomous vehicles: Analyzing Transportation Network Performance and operating scenarios in the Greater Toronto Area. With permission from the authors.)

This graph illustrates both the range of estimates and how these differ depending on the assumptions adopted. Most studies are based on micro-simulations of motorway or freeway traffic with grade separated junctions. In them, the headways that can be safely maintained are critical to capacity estimates. Unconnected AVs may travel with headways of 1.2 seconds (rather than 2 seconds) and under these conditions, a 50% share of the traffic of SAE level 3 may result in a 20% improvement in capacity. Capacity improvements and traffic share seem to vary linearly on most of the range. Connectivity, on the other hand, seems to produce greater and significant improvements in capacity, in particular at 40/50% share of the traffic. In the limit, a fleet of 100% CAVs would treble motorway capacities.

The case of urban roads is more complex. There would be some reduction in incidents and accidents, and improvements in capacity. However, in very congested conditions and without connectivity to traffic signals and other junctions the advantages of CAVs may be smaller. It has even been argued that as CAVs will only perform legal actions (for example, adhere to the speed limit) the performance of urban roads at low levels of CAV penetration may even deteriorate to an extent. This site will try to keep up-to-date on research in all these aspects of CAV deployment. These capacity improvements with minimal or no investments in infrastructure would, in principle, reduce congestion and emissions generating useful time savings.

Below is how to use this type of knowledge to produce a range of outcomes, with links to the time it takes to reach levels of market penetration. It is generally acknowledged that AVs (connected or otherwise) will improve road capacities. It is likely that this impact will be larger on motorways than in roads with at-grade junctions. The latter will probably see greater improvements with CVs. This may result in a perverse situation, where the uptake of (non-connected) automated vehicles increases the use of motorways
(as users look to maximise the benefit of automation) and consequently increases the number of Vehicle Kilometres Travelled (VKT).

4 Reducing the “perceived cost” of travel time

SAE Levels 3 and above permit undertaking other activities while travelling. Level 3 requires the main occupant to be ready to take over at the request of the operating system and this presumably will require the user to be ready to do so at any time. Level 4 removes that requirement provided the vehicle is driven at a permitted ‘driving mode’ on well-marked roads. Finally, Level 5 not only allows complete freedom to undertake other activities, including sleeping, it also permits the vehicle to operate without occupants, for example, to be sent to serve another person.

All of these effects reduce the impact of travel and driving time as ‘lost time’. It will be possible to legally text, read and consult the internet while travelling. The opportunity to undertake these activities, including work, while travelling in a CAV will provide productivity and welfare benefits. This will have an impact on the perceived or Subjective Value of Travel Time Savings (SVTTS) for those using these vehicles. It will not reduce SVTTS to zero, as this is not the case for passengers in taxis and other public vehicles.

Driving is very demanding: holding a conversation with a passenger or listening to an educational podcast, are allowable multi-tasking activities while driving. Texting and reading are not, and holding a hands-free telephone conversation although legal in many countries is recognised as a risky distraction.

Rail and some bus travellers can perform other tasks while travelling like working on a laptop or updating their Facebook profile. Even sleeping is a worthwhile and productive ‘activity’ on a flight for the overworked professional. If instead of focussing on driving, the traveller can perform other useful activities then a reduction in the subjective value placed on shortening the journey would be expected. This would imply a reduction in behavioural SVTTS.

Existing survey tools like Stated Preference or Choice cannot are often used to provide estimates of SVTTS under current conditions. However, these research tools cannot provide a reliable answer to a context and conditions that users will find impossible in practice to envisage. A present estimate of the impact of AVs on values of time must rely, therefore, more on experience and judgment than on behavioural research methods. The prevalent view from the Expert Panel seems to be that there will be a 10% or so reduction in the behavioural SVTTS. That is the parameter in the utility function or generalised cost that multiplies travel time. It has been observed that taxi passengers do not display behaviour consistent with a large reduction in behavioural SVTTS (partly because many taxis charge by time and distance). However, even VIPs driven in limousines at no charge do not seem to have a significantly reduced SVTTS.

The Expert Panel was consulted on this issue for users of levels 4 and 5 AVs. Their prevalent views seem to be that there will be a 10% or so reduction in the behavioural SVTTS that is the parameter in the utility function or generalised cost that multiplies travel time. It has been observed that taxi passengers do not display behaviour consistent with a large reduction in behavioural SVTTS (partly because many taxis charge by time and distance). However, even VIPs driven in limousines at no charge do not seem to have a significantly reducedSVTTS.

It is generally agreed that the Social or Equity value of time of AV travellers should change in line with their Subjective or Behavioural VTTS. This implies also a 20% reduction in the Equity Values of Time of AV users to adopt for the purpose of cost-benefit analysis.

5 New users

The availability of AVs Levels 4 and 5 will allow individuals currently unable or unwilling to drive (because of age, disability or other reasons) to enjoy the flexibility offered by a car.

  • 31% of women in the UK do not hold a driving license
  • 14% men in the UK do not hold a full driving license
  • 46% 17-30 in UK year olds do not hold a full driving license

Some of these new users will opt for owning an AV, others will simply use them as MaaS taking advantage of their low cost and availability. Under certain conditions AV MaaS will provide a better service at a lower cost to society than a low frequency, low use subsidised bus service.

The introduction of these individuals as car (CAV) users is likely to increase Vehicle Kilometres Travelled (VKT), achieving benefits for them while reducing the favourable impact of CAVs on congestion. For example, the project modelling of CAV uptake in Toronto has shown that under certain conditions induced demand will undo between 25% and 33% of the congestion improvements of CAVs.

Level 5 will allow the CAV to travel empty. This will increase VKT compared to a conventional car and create a new range of opportunities and problems. It is here that the CAV mode of use will play a key role. There will be at least three modes of use of AVs: they could be owned by individuals, rented for a period (usually at least a day) or hired to provide a service (a good example of Mobility as a Service, MaaS) as we currently hire a taxi or Uber. Each of these modes of use will have a different impact on traffic and the environment.

An owned CAV Level 5 (and possibly 4) may be used without the need of a parking space at the destination as it could be sent to park elsewhere or simply to keep moving until needed by the owner. It could also be sent to serve the mobility needs of a member of the family or even a friend. This may even lead to a hybrid mode of use, for example, a CAV owner sending the vehicle to operate as MaaS, Airbnb style, when not required (see the ownership model).

6 Parking

AVs Levels 4 and 5 should be able to park themselves without the presence of a driver. This will reduce significantly the demand for parking spaces. In principle, it may be possible for an AV to drive a human to a destination, the human leaves the vehicle and the vehicle finds a parking space or simply drives around until is needed. However, empty AVs adding to congestion is likely to be unwelcome by other drivers.

This reduction in demand for parking spaces may present two opposing effects. On the one hand, it will create new opportunities to use this space to enhance the urban realm and the sense of place in urban areas. On the other, it will reduce revenues from parking and to an extent limit the value of parking policy as a demand management tool. New and better policies will be required to manage demand and traffic conditions and protect social welfare.

The impact on parking will also depend on the mode of use for AVs. AV owners sending their empty vehicle to park or wait elsewhere will create the most difficult problem. This will create additional traffic and congestion, which would annoy drivers or users of other occupied vehicles. Road User Charges (RUC) may be appropriate to deal with this challenge. The problem will be more manageable for the use of AVs as a service as any fleet operator will seek to minimise empty movements.

7 AVs as MaaS

Mobility as a Service can be provided independently from the advent of AVs at any level. However, once AVs are available at Levels 4 and 5 there will be a marked change in their costs and services. The absence of a driver will significantly reduce costs even if the purchase cost of the CAV would be higher. This reduction in costs will encourage some people to abandon the idea of owning a vehicle, automated or otherwise, and choose to use MaaS and public transport as their mobility provider. For those households still owning two or more cars, it may be attractive to own only one and use AV MaaS for the other mobility needs. It is very likely, therefore, that the availability of AV MaaS will tend to reduce the level of car ownership and the overall size of the fleet as MaaS uses it more efficiently serving multiple customers. The perceived cost of using AV MaaS is still likely to be higher than the perceived marginal cost of using an owned vehicle, however, the annual cost of ownership and use will be lower in the MaaS case.

Secondary effects following this shift in mode of use would be:
  • Lower levels of trip induction. Indeed, it is even likely that there will be an increase in public transport trips whenever they are faster/cheaper than using AV MaaS. This would result in a reduction of single occupant trips.
  • The risk of urban sprawl will be reduced and perhaps even reversed as higher density areas are likely to provide a better MaaS service with shorter waiting times.
  • Extensive use of AV as MaaS would be perceived as more equitable. AVs will be no longer viewed as a privilege of the rich.