How will autonomous driving affect infrastructure investment decisions?

Autonomous or driverless vehicles are set to be a transformational technology, with most mainstream car manufacturers having plans to develop partially or fully autonomous vehicles, not to mention Google, Apple and Uber and their ambitions around future mobility, moving away from the traditional model of private vehicle ownership.

Whilst there seems to be lots of discussion on the technical capabilities and issues such as whether the driver will have to be in charge at all times, when it comes to infrastructure, the main focus seems to be on whether or not autonomous vehicles will need dedicated lanes or separated infrastructure or not, but is this really a key issue?

There seems to be a growing consensus that there will not be dedicated infrastructure for autonomous vehicles, and that there will be at least initially a mix of standard and driverless cars. Logically, this has to be the case as the cost of reconstructing large amounts of infrastructure would make a shift to autonomous vehicles untenable in most locations. Besides, the long lead times for planning and construction of infrastructure (which in itself would have a design life of around 60 years) is such that there would be a catch 22 situation where the roll out of autonomous vehicles would be limited as there was no infrastructure, and development of the infrastructure would be limited as demand could not be guaranteed.

So, if autonomous vehicles are not expected to require any specific infrastructure, does that mean there is no immediate issue of concern to infrastructure investors? Well, not really. Whilst there might be some infrastructure considerations that could be put on the back burner for a time, such as developing narrower lanes and strengthening the area beneath the wheel tracks, as driverless vehicles should follow a very precise line (like a rail-less railway), there are investment decisions to be made in the next few years based on enhancing capacity or safety that could be fundamentally flawed if based on ‘conventional’ models of traffic growth.

Pinjari et al (2013) report on studies that cooperative adaptive cruise control can significantly increase highway capacity at moderate to high market penetration, with a 22% increase in vehicles per hour, per lane at 50% penetration rate, rising to a 50% and 80% increase at market penetration rates of 80% and 100% respectively. It should be considered that there are already vehicles on the road now that have adaptive cruise control. They report on another study that for vehicles equipped with automatic braking, sensors and V2V communication could increase highway capacity by 80% at even a 50% presence in the traffic mix. There are an increasing number of cars already in the traffic mix with automatic braking and adaptive cruise control, so even just increasing levels of driver assistance rather than full autonomy can have considerable impacts.

Compare these figures with the capacity added by hard shoulder running of around 25% at peak times, and even as a relatively low-cost option, the seven mile stretch around the M4/M5 interchange around Bristol cost around £90 million and took around two years to complete.

Then consider the cost and timescales for new motorways. The 27 mile M6 toll road near Birmingham opened in 2003 at a cost of £900 million following a three-year construction programme, but had been in various stages of planning and consultation since 1980, a total of 23 years! The proposed new M4 route around Newport has been in discussion since 1991, was dropped on cost basis in 2009, before being resurrected, so 25 years and counting and it remains unclear as to whether it will go ahead or not, and in what form. This 14-mile scheme targeted with improving capacity is estimated to cost £1 billion, if built.

Given that the European Road Transport Research Advisory Council roadmap on connected and automated driving shows deployment of full automation by 2030, that trials of lorry platooning have already taken place and Google and others have suggested they will have a completed project as soon as 2020, one option could be to hold off investment decisions and wait for connected and automated vehicles to increase road capacity on their own.

As with everything however, the devil is in the detail. Davidson and Spinoulas report that whilst automated driving will lead to significant capacity improvements at high levels of autonomous vehicle roll out, at the low levels of market penetration expected in the short to medium term, congestion could actually increase significantly as the increased comfort of autonomous vehicles and potential to engage in other activities will make travelers less sensitive to travel times.

Whilst government organisations will have to wrestle with decisions on whether to invest in increased infrastructure or not, and if so, where and when, they also have it in their power to shape future transport through legal and fiscal measures. This could involve for example promoting the opportunity for shared mobility, where autonomous vehicles link with mass public transport, opening mobility to a far larger percentage of the population, reducing the overall number of vehicles on the road and enabling existing infrastructure to be re-purposed for enhanced walking and cycling space, for planting and improvements to the public realm.

Pinjari, AR; Augustin, B; Menon, N (2013) Highway Capacity Impacts of Autonomous Vehicles: An Assessment
Davidson, P and Spinoulas, A. (2015) Autonomous Vehicles – what could this mean for the future of transport?
ERTRAC Task Force (2015) “Connectivity and Automated Driving”, Automated Driving Roadmap Version 5.0

Martin Lamb, August 2016