There continues to be a lot of hype and excitement around driverless cars, with the first vehicles hitting roads in Britain recently and the NZ Herald running an opinion piece by Paul Minett earlier this week that was generally good, although perhaps a bit excitable about the need to stop all current investments in roads and public transport.
One of the big promises of driverless cars is that they will significantly reduce congestion, as their computer-controlled driving will enable much closer following distances between vehicles, alongside much more efficient operation of intersections. But how will this play out in practice? One of the most detailed pieces of analysis was undertaken by the International Transport Forum (part of the OECD), which modelled in quite a lot of detail what might happen under different scenarios involving the uptake of driverless cars.
Two types of “driverless vehicle” were analysed:
- Taxibots – self-driving cars that can be shared simultaneously by several passengers
- Autovots – self-driving vehicles that pick-up and drop-off single passengers sequentially
The analysis used Lisbon, Portugal as the case study city for the analysis. The different scenarios also looked at whether high-capacity public transport would be available or not, as well as how things would work at 50% and 100% penetration levels of these new vehicles. Some of the results of the analysis are pretty interesting.
Firstly, looking at mode-share, in scenarios where high-capacity public transport is retained the driverless vehicles actuatlly result in an increase in PT mode share, although it seems that they replace all “not high-capacity” PT. This makes a lot of sense, driverless vehicles could make for great first/last mile solutions and for replacing those routes that wind through the suburbs designed primarily to provide coverage. Interestingly walking & cycling mode share is projected to decline from 18% in the baseline scenario to 8% with the new vehicles.
Next, if we look at fleet-size, the projections are pretty sensitive to the different scenarios – varying from a situation where nearly 90% of the private vehicle fleet is no longer required, to other situations where there would actually be more vehicles than the baseline. Once again the existence of high-capacity PT seems to make a big difference to the totals, as does the level of penetration (it seems that most people are expected to hold onto their private vehicles until there’s very high penetration).
Perhaps the most interesting finding relates to projected overall traffic volumes, which increase under all the modelled scenarios (although to very different extents). Scenarios without high-capacity public transport are projected to see substantial increases in car kilometres travelled, from both modal shift away from PT and also the empty “re-positioning” trips taken by the vehicles.
The study highlights that while scenarios with slight increases in travel would be manageable (due to the vehicles themselves being able to travel more efficiently), scenarios with much higher increases are not likely to be manageable at all. Some further detail is provided about the extent of travel increase at different times of the day:
The most interesting trend in the above graph is that the “AutoVots without high-capacity PT” scenario’s greatest increase in vehicle km occurs at peak times, which would be when the transport system is least likely to be able to cope with such an increase. Furthermore, the greatest level of travel increase seems to be on local roads (not motorways), which is probably where we would least want it to happen:
The study then looked a bit closer at where, under the “TaxiBot plus high-capacity PT” scenario, travel increased or decreased. Obviously this would vary depending on the city, but it is interesting to see that most increases are in more peripheral areas rather than central areas. The study itself also highlights that volumes stayed constant or declined on major routes and bottlenecks, with increases mainly confined to local networks (presumably for more local trips?)
Finally, scenarios with full vehicle penetration saw significant reductions in the number of parked vehicles, although once again the reduction was far lower at 50% penetration and actually increased in a couple of scenarios:
There are a few key takeouts from this study that are really important to keep in mind when it comes to discussing driverless cars and how they might change the transport system in the future:
- High capacity public transport remains crucial. Scenarios without high capacity PT saw really big increases in travel demand, especially at peak times. We can rest easy that our current and future rapid transit network investments will continue to provide value in the future – even with a gigantic shift to driverless vehicles.
- Ride-sharing and car-sharing results in very different outcomes. A system based around “car-sharing”, where the driverless vehicles are for individuals, results in a huge amount of travel and large number of re-positioning trips. It also needs a much larger vehicle fleet than ride-sharing.
- All driverless vehicle future suggest a massive reduction in the amount of land required to park vehicles. This could be truly transformational for our urban areas as this land can be repurposed into housing, businesses or open space.
The big take-away though is to note that the introduction of driverless vehicles could play out in a variety of different ways in the future. Some could be really good, others disastrous. It’s pretty important that we get it right.