Audi 25th Hour: Will self-driving cars decongest cities?
Audi has just published its study on congestion, as a part of its 25th hour project: the project started in 2017 and its goal is to assess how self-driving cars will impact the time we spend commuting and the way and the ways in which we operate while commuting.
The published results take the form of a 13-page document. In it, Audi, along with its partners (KIT and Mobility Partners) examine five different scenarios on how traffic could be impacted through the use of autonomous vehicles. Each scenario is based on varying the degree of four parameters: Persons per Vehicle, Innovations in Traffic Management, Distance Travelled per Person, and, of course, Share of Autonomous Vehicles. The experiment was conducted using simulated traffic flows based on Audi’s motherland, Ingolstadt, Germany, and the results make for interesting conclusions.
Autonomy is not enough
Audi examined a “Digital Detox” scenario, which effectively describes a world in which fully autonomous vehicles time travelled from the future and started driving in our cities today. Accordingly, AVs are here but digitalised infrastructure is non-existent, app-based ride-sharing has not progressed, traffic management systems are developing slowly, all while demand for mobility rises due to the higher concentration of people in urban centres. Expectedly, the average number of passengers / vehicle figure is stuck to 1.1, indicating the lack of any improvements in occupancy rates through ride-sharing.
The results are eye-opening: even at AV adoption rates of up to a 60% (i.e. more than half of the cars in the streets are autonomous), travel time and time loss are higher than the current values (by 4% and 12% respectively). With major automakers, such as Ford, promising self-driving cars by 2021, there is an even more alarming finding arising out of this scenario: given the potentially prohibitive cost of autonomous vehicles at the beginning of the autonomous revolution, reaching a 20% AV adoption rate seems like a more likely scenario for the next decade. At a 20% adoption rate and without the requisite modernisation of city infrastructure, the congestion metrics are abysmal: travel time increases by 31% and time loss due to congestion goes up by 51%, compared to today’s levels.
Multi-Modality and Ride-sharing are key
Another scenario examined is “Street Life”. In this case, the average occupancy / vehicle is at 1.5 Persons Per Vehicle, largely driven by the increased uptake of smart ride-sharing and the city’s revolutionised infrastructure which allows efficient transportation management. Importantly, in this high-tech transportation heaven, people choose to walk, travel by bicycle, and use public transportation, which leads to a large decrease in the number of car users - notably, even after 20% of cars are replaced by self-driving vehicles (which traditionally increase the car users, as they allow younger children, the elderly, and disabled people to use cars alone), the number of car users on the streets is down by 17%.
The combination of fewer car passengers and increased car occupancy leads to impressive results, even at very low AV adoption rates. At a 20% AV share, the travel time decreases by 34%, while the time lost due to traffic congestion goes down by 74%. Interestingly, in this scenario, the economic effect of “decreasing marginal utility” is observed: as the percentage of self-driving cars in the network increases, the improvements in travel time and time lost are minimised; this could also be explained by a a concurrent increase in the number of car users, as AVs replace human-driven vehicles.
As the figures show, these astounding improvements in our commuting lifestyle are almost blind to the increased percentage of AVs on the streets. Instead, they are mostly driven by the increase in car occupancy and the decrease in car users.
Autonomous deliveries could be “dangerous”
The last scenario examined by Audi is one stemming from the increasing uptake of online shopping, particularly in urban centres. Unmanned delivery pods could offer a very efficient way of delivering items to buyers round the clock at a minimal cost by taking the driver out of the equation. This could spur a wave of driverless deliveries, and, accordingly, increase the number of vehicles on the streets / decrease the average car occupancy. Although the rise of online shopping would eliminate some trips, such as those to the grocery stores, thus removing some travellers from the system (-10% travellers at the 20% AV adoption level), the increases in travel time and time lost are quite dramatic. With the average occupancy rate down to 0.9 Persons Per Vehicle (from the current figure of 1.1, due to driverless deliveries), the travel time and time lost during commuting would increase by 20% and 37% accordingly, even as autonomous vehicles replace 20% of human-driven vehicles. As the relevant figures show, the lower occupancy rate would only be negated by a replacement of 60% (or more) of our human-driven vehicles - and this would only allow things to return to levels similar to what we experience today (rather than improve). Improvements to the status quo only become evident as almost every car in the streets today is replaced by self-driving cars (80% and 100% rates).
Worryingly, the scenario examined here is likely. Manless delivery pods needn’t be perfect before being deployed: self-driving delivery vehicles, moving at slow speeds in a geofenced area have a lower risk profile and the absence of passengers allows companies to dispatch them without the need to perfect their autonomous technology first. Kroger and Nuro have already partnered up to bring driverless deliveries to the masses, having launched their pilot programme in places like Arizona. Accordingly, it is likely that the growth of self-driving deliveries will outpace the growth of self-driving people transportation.
Do we already have what’s needed?
Autonomous Vehicles can provide solutions to various problems encountered by daily commuters. As Volvo’s 360c concept recently showed, one could reclaim their time by working during their commute, socialising, and replacing their short-haul flying with a self-driving, overnight car trip. What this very first part of Audi’s 25th Hour project reveals, though, is that congestion (or the improvement thereof) is not strictly related to the prevalence of autonomy. Rather, the most dramatic results in decongesting city centres are linked to the removal of cars from the streets and the modernisation of the transportation infrastructure (and technology behind it).
Accordingly, the rolling out of autonomous vehicles without a specific focus on improving the technology behind ride-sharing (through autonomous, on-demand shuttle service, for example) and the urban infrastructure, could actually lead to increased congestion in cities. As various scenarios examined by the 25th Hour project show, lower levels of AV adoption (which, I would argue, would not be exceeded for at least a decade after the first rolling out of autonomous vehicles) could add new car users to the streets without benefitting from the highly efficient use of street space that will become possible after AV adoption reaches higher levels.
Concluding, I would say that Audi’s project is fascinating in that, instead of promoting the idea of an autonomous future, it sheds light on how self-driving cars are not the panacea for congested city centres that many people expect. Most importantly, the 25th Hour makes it abundantly clear that cities need to plan ahead for the arrival of autonomous vehicles: as cities have access to companies driving innovation in the self-driving sphere (through the processes of granting testing permissions, for example), a stronger effort should be made to find ways to adjust cities to the vehicles of the future. The case of dockless scooters and bikes has shown how the lack of preparation can cause disorder in city centres; one can only imagine the kind of upset that could be caused by vehicles multiple times the size of dockless scooters.
Photo Credit: Audi Media