Urban planners should consider how autonomous vehicles can either make cities greener and livable or alienating and more unsustainable. Three phenomena support each other:
- Neighborhoods as walkable urban places organized around mini-downtowns or social hubs,
- Autonomous jitneys to provide first-mile service to transit at the social hubs, and
- Express buses and trains direct between social hubs.
Any two of these could work fine, but together they support each other. This piece explores some of the synergies that the paper Sociable Autonomous Ridesharing introduced. It ends with a series of unanswered questions, and its use of “should” or “could” or “might” or “must” is deliberate. This paper accepts those ideas of livable urban places that have gained a rough consensus: ideas from the New Urbanism, Transit Oriented Development, and other allied sources. It does not explore such concepts as sleeping on the way to work in autonomous pods and trading face-to-face connection for virtual environments. It chews on the problem and the promise of developing social hubs, enhancing transit, and using automated jitneys for the first mile to transit.
A Social Hub’s Anatomy
Each social hub and its neighborhood would have several parts shown on a Google map and explained below:
The social hub
The social hub would include all the buildings and lots right at the crossroads. In addition to a central space, the social hub would have shopping, dining, and entertainment on the ground floor and offices and apartments above. Land use policy should require continuous building facades close to the street – to form the public realm and use land efficiently.
The crossroads would be the intersection of the social hub’s corridors. Traffic at this key intersection would stop completely for buses. Auto-jitneys would stop as close as they could: riders boarding buses and other transit would never be never out of sight of the social hub’s central space.
The social hub’s central space
The central space would be a square, winter garden, or similar public space at the crossroads. The central space would give people a place to enjoy an active public realm. It should have a landmark such as a tower, so people could see it from down the central hub’s corridors. At least some third places (such as bars, restaurants, and coffee houses) should front onto the social space without traffic or parking intervening.
The social hub’s influence area
The influence area would be the land within an easy walk of the social hub – the land from which people are likely to walk rather than ride. Within the influence area, easy access to transit would influence real estate. There, little parking would be necessary, so buildings would be a better use of land. Land-use policy should encourage apartments, offices, and major civic destinations to cluster within the influence area, making it a minor “downtown.” The speed limit in the social hub’s influence area should be below 20 miles per hour, except in the corridors and on any expressways.
The social hub’s corridors
These would be the major thoroughfares – usually arterials but not expressways – shared between social hubs at each end. A social hub’s corridors would extend all the way to the next social hub. Land use policy should encourage infill development along the corridor – especially filling in parking lots and other interruptions in the urban fabric. In the long term, transportation policy should favor reduced noise and pollution in the corridor. This may involve technological improvements such as regenerative braking or a requirement that vehicles run on batteries. Nevertheless, speeds in the portions of corridors outside influence areas could exceed 20 miles per hour.
The social hub’s range
A social hub’s range would be the area within which riders are likely to want to travel directly to the social hub. It would extend all the way to the next social hub. Riders from a social hub should be able to get rides from the social hubs they can get to without passing other social hubs. Residents would probably only identify a quarter of their nearest social hub’s range as being their neighborhood, but the edge is usually fuzzy. The social hub’s auto-jitneys would collect riders from anywhere within the social hub’s range. A ride in an auto-jitney might take 12 minutes to get from one social hub to another. Many of these riders might go from door to door within the range, such as from a house to a school. However, the main use of the auto-jitneys would be door-to-place travel to the social hub’s central space.
Services and Operators
The Transit Agency
A transit agency’s job is to serve a whole city or region as best it can. It must make assumptions about how far people are willing to walk to get to a bus or train. Typically, half of a city’s residents might would be too far away to walk. Even more would be on the wrong side of intimidating thoroughfares. The built environment represents a huge investment that will change slowly. Services such as buses could change quickly, though.
Buses should serve social hubs rather than corridors. In today's transit systems, most buses must crawl along corridors, picking up riders. Their average speed is typically less than 11 miles per hour: every single passenger on a bus must roll to a halt at every bus stop. If auto-jitneys were to pick up riders, though, then the buses would only need to stop every 2-3 miles – perhaps twice as fast as conventionally.
The buses might look and feel much as they do now. They would retain drivers until automation lets them change their to making people feel welcome.
In some ways, buses might be very different, though. They might need more space for bicycles, strollers, shopping trolleys, wheelchairs – and the personal space of riders-by-choice. The next generation of buses with electric motors and big batteries/supercapacitors could look very different from today’s electric buses based on diesel models. In place of an axle, some autonomous electric vehicles use a box with wheels at each end. Motors and batteries fit inside the box. Small low-floor vehicles’ floors are slung between two of these boxes. Larger buses will probably be built with low-floor sections and high-floor sections with batteries underneath. With automation, all the wheels could steer. Buses could then turn very tightly, or even sidle up to a curb diagonally. The buses would also charge themselves along the way – either by catenaries above or by inductive charging from below.
The great benefit of a concentrated influence area around the social hub would be that people traveling to businesses there wouldn’t have to transfer to an auto-jitney to get to their destinations. If commercial development were to concentrate into influence areas, then most commuters would ride an auto-jitney and then transit, not auto-jitney, transit, and auto-jitney again. A secondary, but still critical benefit, would be the synergy developed between workplaces, restaurants, shops, and important institutions. Both of those would help support transit, by making each express stop that much more valuable.
The Local Management
A local nonprofit, business improvement district, or another empowered local unit of government (“local management”) should support the social hub. The social hub should be the interface between the local and the citywide in every sense. Local management would gather land uses that would benefit from for synergies between commercial, residential, retail, and institutional uses. Since the auto jitneys take the place of so many individual driving trips, they would free up parking lots for development.
Curation and development
As auto-jitneys and more effective transit reduce the need for parking, the local management should help intensify the social hub, so that transit there would serve more residents and businesses directly. It should consider moving viable businesses and even well-liked buildings from elsewhere in the social hub’s range into its hub, its influence area, or its corridors. The local management should also manage public places like squares and playgrounds. By caring for the existing building stock and existing residents and businesses, the local management could intensify development without leaving the incumbents behind. Perhaps the biggest key to that strategy is to ensure that while changes might be highly visible (like putting cottages in front yards) they would be applied across the board, creating a coherent, livable new arrangement. Local management should take care to keep retail viable. Today, it is embattled by online competition and often unsustainably indebted. Retail square footage may shrink substantially, but still condense into a tight, synergistic shopping promenade at the social hub.
Except for the stretches of corridors between influence areas, all surface streets should have a speed limit no higher than 20 miles-per-hour. At that speed or lower, auto-jitneys could then mix safely with pedestrians and cyclists. This low-speed environment (and electric cars) could transform sprawl housing. Neighbors might want to cut walkable paseos through long blocks and line them with affordable housing. Cottages might spring up on deep front lawns. Mixed-use paths could carry people walking, riding bicycles, and taking slow auto-jitneys between cul-de-sacs.
Social hubs should be like old-world squares. Each one should have a “mayor” (to borrow William Whyte’s term) to watch over it. It should be compatible with commerce, of course, but its main purpose is to bring residents, businesspeople, and visitors together. Sometimes special events would fill them. Sometimes they would be quiet. Each social hub would be the wider neighborhood’s face to the region, too. Some social hubs might be on one street corner at the crossroads. Others might absorb the crossroads into a shared space. Social hubs must always human-scale, welcoming places.
Local transportation by auto-jitneys would bring people from the social hub’s entire range. The social hub’s street space would be the physical interface between local auto-jitneys and express regional transit. A typical resident might be able to use two, three, or even four auto-jitney services. Curb space at the social hub would be at a premium, partly because competing social hubs could send auto-jitneys to entice riders. Once at a social hub, riders would take express buses and trains to the rest of the region. The auto-jitney service should run mainly door-to-place: service from individual (mainly residential) addresses to the social hub. Suppose a rider today lives about a mile from a bus line and two miles from a social hub. To catch a bus, she would have to walk about 20 minutes to the bus line, and then she would walk to the nearest stop. An auto-jitney could get her to a nearby social hub in about seven minutes, at an average speed of 15 miles per hour (including stops). When the bus came, all traffic would stop, so she would simply cross to the bus from any side – and get on directly. Once on, she would ride faster because the bus would stop all the bus stops the auto-jitneys would have retired.
Auto-jitneys would probably compete with cycling, since they have similar speeds. Both auto-jitneys and transit should make space for bicycles, of course.
Each social hub’s auto jitneys should have a local livery and design. Each social hub should manage (or own and manage) its own auto-jitneys. Inside them, people would see familiar strangers and neighbors they know. They would often go to the same place. At a per-mile cost of perhaps a dollar, a ride on an auto-jitney should be a neighborly public space that happens to be on wheels. Auto-jitneys should be more than just handicap-accessible, too. Riders should be able to roll strollers, shopping trolleys, wheelchairs, and bicycles right onto the vehicle. Once inside, they should also have personal elbow room. The front and back of an auto-jitney should let people know what it “sees” and what it intends to do. Perhaps a green outline in a screen image would let a pedestrian know the vehicle sees her and that it is safe to cross. A cyclist might see himself with an arrow indicating where to pass safely.
Software Services for Auto-Jitneys
For everything to work together, we need different bits of software to talk to each other, but we also need to separate the software into different layers.
At the bottom would be the hardware – the software intimately entwined with the brakes, motors, and other mechanical systems. The others would be the driving, navigating, routing, mobility-as-a-service, social software, and behavior recognition:
The driving system and the navigation system would be separate. The autonomous driving system should be isolated for safety. It must read traffic signs and distinguish vehicles, people, bicycles, obstructions, and road geometry from each other. Then it must drive the safest, most sensible way it could. The driving system must be updated regularly, but since any open connection invites tampering, it should only be updated at a depot, using signed security certificates.
The navigation system would decide which routes the vehicle could take. The navigation system must connect to the outside world – at least to GPS, and probably to emergency updates, such as for weather and closed roads. It must understand signs too – but only for wayfinding. Since security can be broken, the software for driving and safety must be separated from the system for locating a vehicle on the map and finding alternative routes.
The routing system would be concerned with deploying vehicles effectively. While the navigation system could figure out where the vehicle is and which routes it can get to, the routing system would figure out which vehicle should pick up whom and where. The routing system must always communicate with all its vehicles, so it couldn’t be connected to the driving and navigation systems in a way that would make them easy to hack.
Since their driving speed is limited, auto-jitneys’ average speed would be a factor of the number of stops the whole fleet would make and the distance it would travel. At the same time, higher software might let riders know about their fellow riders, and that would influence routing. Suppose that a certain auto-jitney is about to pick up three riders who want to go to the social hub. The second rider lives three houses down on a dead-end street. If the routing system tells her in advance that detouring to her front door will add a couple of minutes’ driving time for the other two, then she may decide to walk to her corner out of curtesy. Providing that information in advance will influence her decision. Of course, if she is frail, then the other two riders would be appalled if she were to walk the distance. It is a human problem, not a technical one. The longer they all live in their homes, the more likely they will be to know each other. They may even look forward to seeing each other, as regular riders on accessible transit often do. The routing system, then, should divulge selected routing details – but not riders’ identities – to fellow riders before they confirm their routes.
The routing system, then, should optimize either the driving distance & time or the number of people the vehicle picks up. It cannot optimize both. The more a vehicle meanders around to pick up riders the longer it will take to get to its destination. Auto-jitneys’ algorithm might optimize for indirect but full auto-jitneys during rush hours and direct, emptier ones during the day and evening. Night service might optimize for direct-to-door service for safety.
Mobility as a service
Software services that coordinate between transit, bicycle sharing, ride sharing, cycling, and walking are called “Mobility as a Service,” or MaaS. Such a system should coordinate with auto-jitneys’ routing software to provide door-to-door service across transportation systems. A single application, then, could book a rider on a combination of three legs: auto-jitney, bus, and another auto-jitney.
The ideal MaaS would let riders optimize their trips for trip-chaining. This is one reason that different auto-jitney services should overlap: so that riders can aren’t captive to the shops at any single social hub. A MaaS for social hubs should include a way to input errands to do – e.g. “Go home and buy a humidifier on the way.” Such a MaaS might make daily commutes and errands less stressful while helping soften peak travel demand.
Social applications would coordinate with MaaS. Suppose a MaaS shows a rider two routes to get to get to a meeting, and it discloses that her brother has already booked a seat on one of them. She might decide to share that leg of her trip with him. Or, perhaps a social application might book five seats together for five friends going to a downtown theater.
It is important that social applications maintain privacy and don’t hurt peoples’ feelings. A very smart MaaS might be able to book those five friends together, but riders might resent that loss of privacy. Moreover, people shouldn’t be forced to define their social connections just to use a MaaS.
One reason that some people hesitate to take transit is that they are worried for their safety. Behavior recognition systems like those being deployed now could help to curb bad behavior and alert authorities. They work by applying machine learning to behavior. A behavior recognition system might recognize a man groping a woman, for instance, and flag the video. The police might verify that is what happened and tag the man for CCTV. They could also make sure the woman can get to safety. Some behavior recognition can tell a dancing street busker from a street fight. Some can recognize emotions. Since a rider could threaten someone quietly, a system that could detect unconscious signs of alarm might save lives without putting victims at risk.
Of course, behavior recognition and surveillance would be an even bigger flashpoint for concern than social applications. It requires exquisite care. Buses would still need someone on board – a “mayor” – to keep it tidy and friendly.
Land Use, Operations, and Livability
Self-driving, express transit, and mobility software themselves are each an incremental change. Yet, one could say that removing the horse from the carriage or installing traffic signals was an incremental change. Incremental changes add up to big changes.
Different streets have different roles
The “Levels of Service” as used by AASHTO and the Highway Capacity Manual are dated on several counts. First, companies like Ford, GM, Daimler, and Tesla are considering ride-sharing and car-sharing models. They are integrating with mobility-as-a-service. Second, a single level of service makes little sense when different modes of transportation have different priorities on different streets. Third, machine learning can help smooth driver/transit/human interaction. Drivers shouldn’t have priority everywhere. If taking the bus is fast because it bypasses congested car/truck traffic, then buses have a higher priority. Instead of a single “level of service” we should discuss which modes have priority on different streets.
One aspect of the different roles is that the corridors' buses or streetcars would be compact for the number of riders they serve, whereas auto-jitneys use more road space per rider. That is, the compact mode goes on the streets most susceptible to congestion, and the side streets less prone to congestion get the less compact mode – although it is still more compact than private cars.
Curb rights – stopping and parking
The local authority should lease stopping and parking spaces like metered parking is today. Auto-jitneys, taxis, ride-share cars, deliveries, bikeshare, and on-street parking should all have their own curb rights. Auto-jitneys should get the prime locations and times, whereas deliveries might be around the corner. Proceeds from curb rights should then subsidize pedestrian & bicycle infrastructure and auto-jitneys.
Social hubs’ traffic
In social hubs’ crossroads, transit and pedestrians should have priority in driving lanes and at the curb. Bikeshare and deliveries should be around the corner. Parking should be at a distance (with free auto-jitney connections instead of nearby handicapped parking).
Pedestrians and cyclists should always feel safe, but some critics worry that they may push their luck or snarl traffic. That is one reason to slow the traffic. If there are sufficient gaps in traffic, then vehicles and probably intersections should alert pedestrians that they have a gap to use. If a pedestrian follows these signals safely, then vehicles must then yield. If too many such interruptions stop traffic, then the vehicles and intersections should signal pedestrians in clumps – until those delays are intolerable, in which case motor traffic should be throttled. Cyclists should mix with all slow traffic in social hubs and ranges. If street lanes fill with enough slow cyclists (e.g. when school is out) speed limits should drop dynamically to match them.
Corridors, by contrast, should prioritize quicker traffic, though always safely. The outer lanes should have lower speed limits than the center lanes. The speed in the center lanes, particularly for transit, should be high enough to get across the city quickly. Corridors would likely maintain bicycle lanes near the curbs (either inboard or outboard of the curb rights spaces.) Crossings, of course, should be monitored for safety and convenience.
Some corridors may look scruffy, with vestigial parking lots and industrial properties. Others may be tightly built-up with apartments and offices, literally shaping a corridor. The more urban they are, the easier they should be to cross.
The streets in influence areas should be slow and pedestrian-friendly. Their curb rights should skew toward on-street parking and deliveries. Most people probably wouldn’t use auto-jitneys in the influence areas since they such a short walk from transit. However, some people would want to get elsewhere in the range, and some would want to get to adjacent social hubs.
Speeds on streets or roads in the ranges would be 20-miles-per-hour or less. Sometimes, neighbors may agree to build woonerfs or even pedestrian passages that don’t even allow auto-jitneys. Perhaps some side streets and roads can be narrowed, but many can provide on-street parking to free up land on lots for development. Even if people drive far less, the remaining curb space may be handy. The on-street parking should be priced low enough to keep it worthwhile to replace off-street parking with newly-built square footage.
Land Use Planning
Density & intensity
In principle, land uses should peak in intensity toward the social hub and peter out in-between the social hubs. If we think of density as a mountain, with the highest density being the peak, then it should be highest at the crossroads – or at a rail station combined with the social hub. The smallest blocks should be around the social hub, with larger ones farther away.
Social hubs’ and influence areas’ land use
In practice, the mountain may be more like a volcano with a caldera instead of a tip. Old crossroads typically have old and beloved low-density buildings. Older buildings can also be perfect for incubating businesses. In practice, the local management may need to transfer development rights from the buildings in the social hub to the influence area. In general, the frontage directly facing social hubs should be more public – less local – than the side streets in the influence area: hot bars in the social hub; neighborhood joints in the influence area. Retail catchments would be roughly a quarter of the size of their ranges, since the ranges extend all the way to the next social hubs.
Local management should move destinations that transit riders are likely to frequent into the influence area. Moving a 15-person office from a mile away could save auto-jitneys about 348 seat-hours of travel time per year – assuming they would instead walk five minutes than take the auto-jitney. Since the local management would probably have to subsidize those hours, the effort to entice such an office toward transit could well pay off.
The smallest blocks should be toward the social hub. It is the most significant destination. Since curb rights are precious and the streets wouldn’t be noxious, landowners or the local management might choose to cut narrow streets through empty lots to the next block. Lobbies to new apartment buildings on street corners might face the side street to keep retail frontage free.
Social hubs and influence areas could also collect old buildings – especially retail frontages – moved from the corridors.
Corridors between social hubs would vary in character. Most planners and urban designers would probably prefer them lined with tight, space-defining buildings. This planning ideal usually includes continuous retail all along the strip, but it’s unlikely that demand would be sufficient to line the whole distance. In a social-hub model, too, transit just shoots through the corridor to the next social hub. Corridors would have the advantage of easy vehicular access, but not of direct transit access. While corridors’ roles would vary, most would probably favor either residential uses like apartments or light industrial uses that need vehicular access more than transit access. A cabinet maker with a showroom, for instance, might want easy road access might also want to put his store as close as possible to the social hub. The city or town should certainly limit auto-oriented businesses – except for what might pass for their quaintness in a generation.
The side streets and back streets of a social hub’s range may also increase in density. If the influence area has walk-up apartments already, it might be difficult to build much more density there. However, a range consisting mainly of single-family houses on large lots could absorb plenty of development. A homeowner who builds a cottage on the front lawn and a granny-flat in the back yard could easily triple the number of units on her lot – and add an office in an attached garage too. Since the ranges make up the bulk of land, they may represent the most capacity for infill development. Leafy residential lots might have even more to gain from auto-jitney service than property in the social hubs.
Readers may object that this is all fanciful, but it’s falling into place. Consider these trends:
- Auto-jitneys are already deployed in pilots all over the country. The first places using them are campuses where the driving problem is simpler. Sometimes they mix with traffic and pedestrians.
- Transit agencies are speculating that “last mile” or “first mile” service can let transit operate more efficiently. Some agencies run simulations of how reducing the number of stops they have to service would improve service.
- Walkability and livability are now mainstream. Developers seek out physical environments where they can build such value.
- Local business improvement districts and other institutions help coordinate urban development, fund infrastructure, and make places work better.
- Cities that face high housing costs, or just high costs in gentrified livable places, are looking for ways to increase development, while neighbors worry about excessively tall buildings and additional traffic. Auto-jitneys and transit that obviate parking can free up land without generating additional traffic.
This vision is well-grounded. Projects like sAVe, an entry for a competition from Blank Space New York, have proposed using auto-jitneys in almost the form discussed here. The real questions are these:
- What combination of software and physical design will help riders feel comfortable riding with strangers? Having sufficient personal space, assistance available for riders, and riders’ familiarity with fellow riders should help. What else?
- What improvements to public space will help people look forward to changing between auto-jitneys and transit at social hubs? What kinds of information would help (i.e. knowing whether your friends are there.)
- How can cities and local management develop the backbone to help concentrate land-use into livable (and not oppressively overdeveloped) influence areas around transit stops? What would impart the discipline to keep them livable?
- How many auto-jitneys would be needed to service ranges of several square miles? What would make them viable for doing errands and having fun, not just commuting?
- How would an all-express transit system work? How could it be optimally effective?
- How could cities subsidize auto-jitneys in areas with lower incomes? Could a consortium of businesses support mobility so that employees can ride? How could local management keep the auto-jitneys running, safe, clean, and crime-free? More, how could riders without smartphones (or the future equivalent) use them, and what sort of account would people without debit or credit cards use?
- How can local management help keep local businesses viable despite what may be higher costs? Although the social-hub model would apply across whole cities, it’s possible that pricing might mimic gentrification.
These are all “how” and “what” questions, rather than “will” or “can” questions. These questions are just as solvable as the question, “But where will people park” was in 1930. The biggest question, then, is when do we start?