Go up one level Go home

1 Introduction

There are two types of challenges to the introduction of CAVs. There are the challenges of the technology itself: ensuring it will it work at all times, overcoming difficulties with lower levels where the user must be able to take control of the vehicle at certain times, and so on. The other challenge is the set of potential disadvantages, or social costs, associated with the use of CAVs, in particular at the higher levels of automation. This includes trip induction, unemployment and negative impact on public transport.

Traffic accident risks are reduced with CAVs, but some new risks may emerge. Among them, the risks of:
  • Technical failure resulting in a different type of accident
  • Misuse of a vehicle for criminal purposes, including terrorism
  • Hacking of personal data
  • Demands on critical journeys

There is also the risk that CAVs, in making longer journeys easier, could induce urban sprawl that would be increasingly difficult to serve by public transport. This, of course, will depend on land use regulations and enforcement.

We discuss first technology and human interface risks and then those aspects affecting the impact on society.

2 Technology Risk


Highly Automated Vehicles (HAVs) are still at the prototype stage. There is little evidence of how reliable they are in mixed traffic and in different scenarios. One of the main challenges is proving software safety, known as 'verification and validation' or V&V. Software for automated driving must be designed and developed too much higher standards from that currently found in consumer devices. Engineers need new methods for designing software that can be proved correct and safe in complex and rapidly changing conditions. Current ADAS features (up to Level 3) rely on the driver taking control in the event of an anomaly. This means they can err on the side of caution against positive actuation, for instance, the system can fail to take action in response to an incident and canal low the human driver to override.

The ongoing public road testing in California is one of the most comprehensive sources of information on AV performance. This is because, under the California DMV rules, every company that has a state permit to test AVs in California must report how many miles they have driven in automated mode and how many times a driver/safety operator had to intervene. The California DMV published its second annual disengagement report in February 2017. The DMV rules define disengagements as deactivations of the autonomous mode in two situations:

(1) "when a failure of the autonomous technology is detected," or
(2) "when the safe operation of the vehicle requires that the autonomous vehicle test driver disengage the autonomous mode and take immediate manual control of the vehicle."

In adopting this definition, the DMV noted: "This clarification is necessary to ensure that manufacturers are not reporting each common or routine disengagement."

The report includes how many miles each company’s cars have driven and how many times a human has had to take back control – an event that is called a disengagement.
Company
Miles
Disengagements
Miles / Disengagements
Waymo/Fiat Chrysler
635,868
124
5128.0
Cruise (GM)
10,015
284
35.3
Nissan
4,099
28
146.4
Delphi
3,125
178
17.6
Bosch
983
1,442
0.7
Mercedes-Benz
674
336
2.0
BMW
638
1
638.0
Ford
590
3
196.7
Tesla Motors
550
182
3.0
2016 Annual number of reported AV testing miles and disengagements in California
The performance metrics (number of miles and number of disengagements) could be considered an imperfect measure of how the technology performs. Each company is at a different stage in its development of automated driving systems. They have performed different levels of testing on various types of public roads and in differing conditions. The metrics also do not take account of other types of testing, such as on private test facilities or in simulators. It is, therefore, difficult to make comparisons between companies. Further commentary on these metrics is provided in this article by Wired.

As shown in the table, Waymo, the Alphabet subsidiary, drove more than 635,000miles in California last year, during which time the human took back control of the vehicle 124 times, which equated to a disengagement rate of 5,128 miles per disengagement. According to Waymo’s reports, this could be seen as a substantial improvement in the technology over the results from 2105. During that year, they drove 424,000 miles and disengaged 341 times. The causes for disengagements are listed as:

  • Software discrepancy
  • Weather conditions during testing
  • Recklessly behaving road user
  • Unwanted manoeuvre of the vehicle
  • Perception discrepancy
  • Incorrectbehaviourpredictionofothertrafficparticipants
  • Construction zone during testing
  • Emergency vehicle during testing
  • Debris in the roadway

A more recent report from Waymo states that its vehicles have covered more than 3.5 million miles on open roads. The report details the challenges of testing these vehicles and how the company is trying to address them.

It is clear that all HAV developers are still some years away from having Level 4/5 technology that can be allowed on public roads.

3 User-Related Issues

There will be very few requirements made of a Level 4/5 AV user. They should be able to summon a vehicle and board it, specify the destination in unambiguous terms (perhaps helped by the software), and pay any charges due. Presumably, the vehicle itself will take care not to stray beyond its ‘valid operational domain’ in the case of Level 4. However, for lower levels of automation, or for hybrid cases where a Level 4 must be human-driven outside its operational domain, other user-related issues arise. These include:

  • Boredom – could lead to distraction
  • Behaviour-related issues – includes reallocation of attention when required
  • Resuming control – also known as the ‘handover problem’, this issue is about whether the user will react quickly and reliably enough
  • Situational awareness – can be significantly reduced by automation and this may be a problem in the case of an incident
  • ‘Out-of-the-loop’ performance – as a result of reduced situation awareness (and other factors mentioned) performance can drop when the driver has been out-of-the-loop
  • Complacency – operators could attend to manual tasks at the expense of the automated task

A much longer and comprehensive list of user-related issues can be found in a recent publication on the Evaluation of Intelligent Road Transport Systems.

4 Trip Induction

There are several reasons to think that CAVs may generate more trips than a fleet of conventional cars. For instance, new users among those unable to drive, unoccupied movements, and the greater convenience of using a vehicle without having to find and pay for a parking space. The extent of trip induction will also depend on the mode of use of CAVs. The key issue here is likely to be the different perception of costs between owning a car and hiring one.

Car owners have only a vague understanding of the costs of using their vehicles. The best approximation would be a perception of fuel costs per kilometre, plus tolls and parking costs. This is an approximation of their true marginal costs. Maintenance, insurance, road tax and depreciation are often seen as semi-fixed costs of owning the car and not directly related to driving it. In contrast, when hiring an Uber or a taxi, all costs, including the driver, must be borne and these are made very clear in the fare payable. This imperfect perception of costs favours ownership over hiring any kind of vehicle and the greater convenience of CAVs is likely to generate more trips and VKT among owners.

The trade-off will be different for those who use CAVs as MaaS instead of owning them. There is some evidence that those who have replaced their own vehicle for membership of a car club reduce the number of VKT per year. This is partly due to the direct perception of full costs and partly to the slightly onerous task of reserving a suitably nearby vehicle.

CAVs for hire as MaaS will serve several customers a day. So the fleet necessary to deliver these services would be smaller than if they were owned and remained stationary most of the day. A smaller fleet is cheaper to provide, releasing resources for other uses, and although CAVs will be replaced more often (a cost) they will be more modern (a benefit).

Therefore, it is likely that the extent of trip induction will depend on whether the majority of CAVs will be owned or hired. This is very difficult to determine from current perspectives. The expert panel estimated that if CAVs are owned they will probably generate some 10% more VKT. The same panel estimated that if CAVs are used as MaaS, then there might be a 10% reduction in VKT compared with the use of a conventional car.

5 Unemployment

CAVs at Level 5 (and to a lesser extent Level 4) remove the need for a driver in taxis, minicabs and Uber/Lyft-like services. Trucks and buses will also reach Level 5 and will have a similar impact on employment. This will not only affect drivers, but related professions, such as driving instructors and examiners, vehicle insurance personnel, parking assistants and, eventually, even traffic police. Some may be redeployed as CAV fleet managers and risk assessors, but the task of retraining the majority of other professions will be a major undertaking. This impact will be greater when the proportion of Level 5 vehicles that are for hire is high. This is because MaaS operations are likely to adopt Level 5 faster than the individual ownership market will, thus accelerating the problem.

6 Liability

This is a difficult issue and has not been settled. With SAE Level 2, it is clear that the situation is not different from that of a conventional car. The driver is responsible for manoeuvres, the manufacturer for the vehicle, and the owner for keeping it roadworthy. Level 3 transfers some liability to the manufacturer, as the vehicle must alert the driver to take over. With Level 5, the occupant is no more liable than a passenger in a taxi or bus, unless they are the owner, in which case they must make sure the vehicle is roadworthy. The same applies to Level 4, provided the vehicle is used in the correct driving mode.

It is therefore likely that car insurance will be displaced from the driver to the vehicle and may eventually be retained by the manufacturer, as is the case with washing machines. For failure with proper use, the liability lies with the manufacturer, whereas for failure with improper use the liability lies with the owner. This will mean a reduction in insurance premiums from Level 2 to Level 5 with the possibility of zero premiums for the owner unless they retain the possibility of taking over at will.

7 Environment

The potential impacts of CAVs on the environment are challenging to predict with confidence, due to the uncertainty around timescales, technology and uptake. The environmental impact is highly correlated with the consumption of fuel and energy. For example, vehicle connectivity and automation could start to impact the following positively:

  • Congestion mitigation – improving traffic flow and reducing accident frequency, which is a major source of congestion
  • Eco-driving –reducing fast acceleration and breaking
  • Platooning –reducing aerodynamic drag from vehicles traveling closely, especially on highways
  • De-emphasised performance – if drivers become passengers, this may reduce demand for vehicles that are capable of high rate of acceleration
  • Improved crash avoidance could lead to less need for vehicle crash-worthiness and lead to lighter vehicles
  • ‘Right-sizing of vehicles’ –passengers could order an AV that is the right size for a single journey, rather than purchasing a vehicle that can accommodate all possible journeys

Automation could also have the following negative impacts on the environment:

Increased travel due to reduced costs of travelling. This could include financial costs (for example, from lower insurance premiums), the perceived discomfort costs of driving and the attractiveness of using that time for other purposes.
Those that are unable to drive due to age, disability, intoxication or lack of car availability or ownership will be able to ‘drive’.

As an example, a paper by MacKenzie, Wadud & Leiby attempted to estimate the energy impacts of AVs in the US. The complete answer has to take into account whether CAVs will induce more VKT and resulting congestion, or reduce VKT when used as MaaS. How this balance will materialise in different countries and regions is uncertain at this stage.

8 Impact on public transport and active modes

AVs could be ideal feeders to public transport services, particularly where they offer an advantage over driving. They could also persuade some people that the convenience of a CAV justifies its use for the whole journey. The effect of CAVs on public transport will depend on a number of issues, such as the quality and frequency of the public transport service and its protection from congestion. For example, segregated fixed-track services. The impact of CAVs on the number of people using active modes is as yet unknown.

Some of the induced trips will be abstracted from current public transport patronage, in particular from low density, low-frequency buses. Most research agrees that CAVs, used as MaaS, will provide an attractive alternative to bus services in many areas. There are not many estimates of this loss of patronage because it would be heavily dependent on market share, route density, frequency and how available MaaS, with or without ridesharing, turns out to be. On the other hand, it is believed that high-performance public transport services that have a protected right of way and are not delayed by congestion will be less affected.

Given the levels of uncertainty about these impacts, the expert panel was consulted about them for the initial levels of penetration in traffic(10%and20%)and for bus and fixed track(or protected right of way) services separately. In the case of bus demand, the panel average expectation is a drop demand of around 10% (for CAV 10%) with a marginally greater one for 20% CAV penetration. These figures hide regional variations. In the US&Canada, the reduction is closer to 30%. In the rest of the world(including Europe, Australasia and Latin America), the decline in demand is minimal, perhaps reflecting greater public transport mode share, coverage and frequency.

Only low CAV penetration of up to 20% of the fleet was considered in the Delphi poll

Pic 18.png
Pic 18.png


Looking at the chart, it is apparent that some respondents expect AVs to contribute to bus demand. The comments provide some suggestion for the variety of responses. It is argued that the use of hired AVs will reduce car ownership and that a bus will often be almost as good as an AV, and possibly cheaper. This argument will vary by city and even by where in the city one resides. It is likely that low-frequency bus routes will suffer most, as they are easily replaced by AVs. Therefore, it will be important to adapt these results to each local context. On average, the expectation was a 10% loss of patronage at 20% CAV market share.

In the case of fixed track public transport, the panel expected its protection from congestion to help it remain attractive in areas of high traffic volume. As with buses, any estimate of the impact of CAVs is limited by wide regional variations, public transport market share and other factors.

The panel respondents agree that fixed track with a protected right of way will suffer a smaller reduction in demand than buses. This is partly because of the greater speed and insulation from congestion and partly because they are only justified when demand is high and therefore frequency, a defence against CAV availability is also high.

Fixed track systems also seem to fit better with CAVs as feeder modes, now that no large ‘Park & Ride’ parking spaces are required. This explains why some respondents expected fixed track services to gain demand with CAVs.

Pic 19.png
Pic 19.png


There remain many unknowns in establishing a sensible expectation for the impact of CAVs on public transport. On average, there was no expectation of change at 10% CAV market share and a 10% loss of patronage at 20% CAV penetration.

Category
Abnormal / Challenging Driving Event
Issues involved
Obstructions
Parked vehicles
How to ensure they are parked and have not momentarily stopped. How to allow for possibility of doors opening?
Disabled (broken down / crashed) vehicles
Passing may lead to compromising other rules of road such as crossing solid white lines
Pedestrians
How much space to leave? Different clearances for different types? Pedestrian behaviour can be unpredictable. Should vehicle slow down when passing pedestrians on footway?
Passing cyclists
How much space to leave? Different clearances for different types? Cyclist behaviour can be unpredictable.
Road flooding
Difficult to sense the depth? Could lead to loss of control of vehicle or splashing of pedestrians.
Animals in road (either shepherded or loose)
For smaller animals, it can be difficult to decide whether to pass over animal or attempt to stop or swerve.
Ridden horses
Determining appropriate speed and overtaking strategy.
Negative obstructions such as potholes or road/bridge collapse
Could be difficult to sense.
Load shedding from other vehicles
Action could depend on density/mass of objects being shed, but might not be possible for machine to determine.
Vehicles in process of becoming disabled, e.g. tyre blow out, lorry jackknifing, tall vehicle overturning in wind etc.
Challenging for machine to detect and interpret subtle clues that provide indications.
Traffic calming measures
Speed humps, chicanes, etc. need to be detected and negotiated appropriately.
Fallen power cables/branches in carriageway
Could be challenging to detect with sensors.
Level crossing
Similar to traffic signal but consequences of stopping on rail line could be catastrophic.
Overtaking
Challenging to detect oncoming vehicles
Lane reallocation / rerouting
Temporary lane closure on highway
Road layout may differ from map being referred to by vehicle
Temporary contraflow
Automated driving feature may be designed for highways and not for two-way traffic operation
Lane designations (e.g. bus lanes, high occupancy vehicle lane, hard shoulders)
May need to clarify under what circumstances vehicles can enter.
Adverse weather / environmental conditions
High winds
Loss of control
Snow either falling or on carriageway
Sensor visibility compromised, road markings and kerb lines obscured, loss of vehicle control
Heavy rain
Sensor visibility compromised, road markings obscured, loss of vehicle control
Ice
Loss of control
Fog / bright sunshine etc.
Sensor visibility compromised
Road Etiquette
Emergency vehicle in vicinity
How to avoid impeding whilst obeying rules of road
Crossing white lines
Under what circumstances can vehicle do this?
Interpreting gestures from other road users
Challenging to detect and interpret the meaning of hand gestures, flashing of headlights, etc.
Traffic flow arbitration
Police or authorised persons intend to stop AV
How does AV recognise what is an authorised person, and then interpret commands?
Two lanes merge into one
Often involves interaction between human drivers
Merging onto highway
Often involves interaction between human drivers
T junction
Poor lateral field of view from AV
Cross-roads
Poor lateral field of view, right turn stale-mate
Temporary speed limits
How to ensure location is communicated to AVs
Temporary traffic signals
How to ensure location is communicated to AVs
Temporary stop-go sign
Can AV interpret?
Giving way to oncoming vehicles on narrow section of road
Often required communication between human drivers to decide who proceeds first
Roundabouts
Detecting correct lane allocations, ‘give way to right’ standoff
Zebra crossings
Giving way to waiting pedestrians
Traffic signal failure
Junction reverts to priority based or interaction between human drivers