GM investing $5.4B in US plant improvements over next 3 years
Tesla launches line of Li-ion stationary storage systems for homes and businesses: POWERWALL and POWERPACK

ITF report finds self-driving shared vehicles could take up to 90% of cars off city streets; total kilometers travelled increases

A fleet of self-driving shared cars combined with high-capacity public transport could make 90% of conventional cars in mid-sized cities superfluous under certain circumstances, according to a study published by the International Transport Forum (ITF) at the OECD. Even during peak hours, only about one-third (35%) of the current number of cars would be needed to provide the same number of trips as today.

However, while the number of cars is drastically lower, total vehicle kilometers travelled (VKT) increase—more than doubling in one scenario at peak periods due to detours for pick-ups/drop-offs, repositioning and a shift from bus trips to shared cars. The additional travel could increase environmental impacts, if the fleets used conventional engines. If a fleet of electric vehicles were used instead, a fleet of shared self-driving vehicles would need only 2% more vehicles, however, to accommodate battery re-charging times and reduced travel range.

ITF researchers used actual transport data from Lisbon, Portugal, to modal the impact of two concepts: “TaxiBots”, self-driving vehicles shared simultaneously by several passengers (ride sharing) and “AutoVots”, which pick-up and drop-off single passengers sequentially (car sharing).

Other findings of the study include:

  • A TaxiBot system with high-capacity public transport will result in 6% more car-kilometers travelled than today, because these services would have to replace not only those provided by private cars and traditional taxis but also all those provided by buses. An AutoVot system in the absence of high-capacity public transport will nearly double (+89%) car-kilometers travelled. This is due to repositioning and servicing trips that would otherwise have been carried out by public transport.

  • A TaxiBot system in combination with high-capacity public transport uses 65% fewer vehicles during peak hours. An AutoVots system without public transport would still remove 23% of the cars used today at peak hours. However, overall vehicle-kilometers travelled during peak periods would increase in comparison to today. For the TaxiBot with high-capacity public transport scenario, this increase is relatively low (9%). For the AutoVot car sharing without high capacity public transport scenario, the increase is significant (103%). While the former remains manageable, the latter would not be.

    Spatial distribution of the variation of peak hour travel volumes for TaxiBot system in Lisbon. (weekday 8-9 a.m., TaxiBot plus high-capacity public transport scenario, vehicle-kilometers). Source: ITF. Click to enlarge.
  • In all cases examined, self-driving fleets completely remove the need for on-street parking. This is a significant amount of space, equivalent to 210 soccer fields or nearly 20% of the curb-to-curb street space in the model city. Additionally, up to 80% of off-street parking could be removed, generating new opportunities for alternative uses of this valuable space.

  • An AutoVot fleet requires more vehicles than a TaxiBot system to provide the same level of mobility. AutoVots also require considerably more repositioning travel to deliver that mobility. Around 18% more TaxiBots and 26% more AutoVots are needed in scenarios without high-capacity public transport, compared to scenarios where shared self-driving vehicles are deployed alongside high-capacity public transport. Without public transport, 5,000 additional cars are required for the TaxiBot system and another 12,000 in the AutoVot system. Car-kilometers travelled would increase by 13% and 24% respectively.

  • If only 50% of car travel is carried out by shared self-driving vehicles and the remainder by traditional cars, total vehicle travel will increase between 30% and 90%. This holds true irrespective of the availability of high-capacity public transport. Looking only at traffic during peak hours, the overall number of cars required increases in all but one scenario, namely TaxiBots with high-capacity public transport.

From the results, the ITF researchers posited a number of policy insights:

  • Self-driving vehicles could change public transport as we currently know it. For small and medium-sized cities it is conceivable that a shared fleet of self-driving vehicles could completely obviate the need for traditional public transport.

  • The potential impact of self-driving shared fleets on urban mobility is significant. It will be shaped by policy choices and deployment options Transport policies can influence the type and size of the fleet, the mix between public transport and shared vehicles, and ultimately, the amount of car travel, congestion and emissions in the city.

  • Active management is needed to lock in the benefits of freed space. Shared vehicle fleets free up significant amounts of space in a city. Prior experience indicates that this space must be proactively managed in order to ensure these benefits are fully realized. Management strategies can include restricting access to this space by allocating it to specified commercial or recreational uses, such as delivery bays, bicycle tracks or enlarged footpaths. Freed-up space in off-street parking could be used for urban logistics purposes, such as distribution centers.

  • Improvements in road safety are almost certain; environmental benefits will depend on vehicle technology. The deployment of large-scale self-driving vehicle fleets will likely reduce both the number of crashes and crash severity, despite increases in overall levels of car travel. Environmental impacts remain tied to per-kilometer emissions and thus will be dependent on the adoption of more fuel-efficient and less polluting technologies. TaxiBots and AutoVots are in use 12 hours and travel nearly 200 kilometers per day, compared to 50 minutes and 30 kilometers for privately-owned cars today. More intense use means shorter vehicle lifecycles and thus quicker adoption of new, cleaner technologies across the car fleet.

  • New vehicle types and business models will be required. A drastic reduction in the number of cars needed would significantly impact car manufacturer business models. New services will develop under these conditions, but it is unclear who will manage them and how they will be monetized. The role of authorities, both regulatory and fiscal, will be important in guiding developments or potentially maintaining market barriers. Innovative maintenance programs could be part of the monetization package developed for these services.

  • Public transport, taxi operations and urban transport governance will have to adapt. Shared self-driving car fleets will directly compete with urban taxi and public transport services, as currently organized. Such fleets might effectively become a new form of low capacity, high quality public transport. This is likely to cause significant labor issues, ITF suggests. Governance of transport services, including concession rules and arrangements, will have to adapt.

  • Mixing fleets of shared self-driving vehicles and privately-owned cars will not deliver the same benefits as a full TaxiBot/AutoVot fleet—but still remains attractive. In all fleet-mixing scenarios, overall vehicle travel will be higher. Also, vehicle numbers will increase in three out of four peak hour scenarios. Improved traffic flow of automated cars could mitigate congestion up to a point. However, the public policy case for self-driving fleets alone (without high-capacity public transport) may be difficult to make based solely on space and congestion benefits, due to the increase in overall travel volumes. Nonetheless, even in mixed scenarios, shared self-driving fleets could be a cost- effective alternative to traditional forms of public transport, if the impacts of additional travel are mitigated. “All in” deployment of shared self-driving fleets may be easier in circumscribed areas such as business parks, campuses, islands, as well as in cities with low motorization rates.

The “Urban Mobility System Upgrade: How self-driving cars could change city traffic” project has been carried out with the Corporate Partnership Board of the International Transport Forum.




A fleet of AWD autonomous drive e-mini-buses (4 to 10 passengers) with appropriate call/control system could pick-up passengers at/near home and take them to e-buses and/or e-trains main routes or directly where they have to go.

A well managed system could certainly reduce private car usage, pollution, GHG, travel time, accidents and energy used.

This could very well be the solution to reduce traffic jams and every increasing number of vehicles on our roads and streets.


You have to have sharing to get the congestion reduction benefits.
If you can get people to go to bus stops, or even bus routes, you gain another advantage. If you expect every shared taxi to go to your door, you have a problem.
If it can drop you to within 200 metres of your destination, it is much easier to implement.
You could think of it as having bus stops at twice or 3x the usual frequency along a route.
You would be crazy not to mix in some high capacity public transport as well, even 80 seat buses for mid sized cities - larger cities might need trains.


Was on a bus through the town centre that took a circuitous route with frequent stops to reduce the distance walked by passengers - and increasing the trip time. Thought something like TaxiBots would solve the performance compromise. Good to see some simulations on the subject.

A driving factor for me would be waiting time - frustrating to wait 10 minutes if you are only 2km from destination. The drop-off points could be adaptive. In non-peak times the Taxis might be able to go to the door, whilst during peak times they might should a least-bad route that gets them back to the next pick up point in time.


People buy cars for personal transport but the type of car they buy is influenced by their personality. This is why we see fuel wasting Hummers, etc. on the road. A shared car on the other hand can be the smallest, most energy efficient vehicle available. My city's Car2Go program uses Smart cars for example. If you can take 100 30mpg cars off the road and replace them with 65 80mpg ones, that has to be a good thing even if you're driving twice as many miles. Using them to solve the "last mile" problem for mass transit systems would be even better.


This eliminates those "fat, lazy, overpaid, union" bus drivers Harvey hates.


Well said SJC.

We no longer need elevator and bus operators-drivers.

Future autonomous (on-demand) 4 to 10 passenger e-vehicles could replace the majority of current polluting private cars and city buses.

Since it may take 20+ years to introduce, current drivers will have time to be retired and/or retrained to do something else.


My take away is this...

If 'taxibots' exist, then its like carpooling. 1 car versus 2-3 cars for the same trip.

If the number of on the road vehicles drop by 33-90% this frees up millions if not billions of hours that could be spent on something productive instead of traffic.(that and traffic is a waste of fuel)

Again, safety/efficiency, odds are computers can be better drivers than any human. Especially if they are connected to infrastructure and know the timing of lights and what is laying in wait over a crest or blind spot.

As others pointed out, this could give rise to really efficient cars. If something like a model S could be bot driven, it could paid for by fares really quickly. So it could justify the extra expense of being green or better for the environment. (if these had a 5-10 year service life, we could be looking at 182,000-732,000miles).

Again as I said in the past, it would really be easy to encourage ride sharing/carpooling with autonomous vehicles. If carpooling was incentivized with significant discounts it would become much more popular.

Vehicle miles will probably increase, but I don't see it being double, maybe a 33% increase at the most. The double mileage might come from a perceived population growth, or opportunity to travel by car where there might not have been one in the past. I think 33% per capita increase in miles would be a lot in the USA, double would be a stretch. Like I said odds are ride sharing will drop a lot of the extra miles back down to nominal levels.

The comments to this entry are closed.