Will Autonomous Vehicles Usher in a Spotless Future?


There is no shortage of speculation about what kind of impact autonomous vehicles will have on cities. One commonly touted theory is that they will significantly reduce the need for parking lotsfreeing up an enormous amount of land for redevelopment. This new land could provide a huge opportunity for cities to mitigate other pressing issues such as housing affordability, economic development, and park access. 

“If parking lots become apartments, 2.5 million housing units of housing would be added to NYC”

For instance, if all parking lots are converted to parks, there would be 8 thousand acres of park added to NYC, or almost 10 more Central Parks. If parking lots become apartments, 2.5 million housing units of housing (at 1,000 sf per unit) would be added to NYC, a 75% increase in the number of housing units! Or, if parking becomes office buildings, that space could house 12.42 million jobs. Understanding this immense potential should inform how we envision the future of cities with autonomous vehicles. Because of the enormous impact this could have on cities, KPF initiated a research project to both quantify the off-street parking in New York City, as well as the simulate impact of redeveloping it.  


The degree to which existing parking will become unnecessary, if at all, is very much up for debate. Discussions about this topic explore a range of scenarios about how much parking will be reduced, and where within a city those reductions will take place. Because of this uncertainty, we approached this research project with the goal of avoiding a single deterministic outcome. Instead, we embraced uncertainty so that we, and everyone else, could explore a large range of potential futures. 

Our solution was to build an interactive web app, called Spotless, that put’s the user in the driver’s seat. Recently unveiled at the Metropolis Think Tank panel Leveraging Technology for Adaptable City Planning, this app can be used to explore a large range of potential redevelopment scenarios by deciding what amount of parking lots get redeveloped, and what they get redeveloped as. Through the use of a single slider, they can determine percentage of parking lots that: 

1) stay as parking lots 

2) get converted to park 

3) get redeveloped as residential use 

4) get redeveloped as commercial/office use 


This allows them to explore extreme scenarios, such as converting every parking lot to park space. New York would become a much greener city with more immediate access to parks in many neighborhoods. Or they can explore one of a countless number of mixed scenarios. Perhaps 25% of parking lots become parks, 50% become residential, and the remaining 25% stay as parking lots.  

We categorized all of NYC’s off-street parking into 4 types.

We categorized all of NYC’s off-street parking into 4 types.


In order to create these scenarios, we first had to find every parking lot and garage in New York City. To do this we took advantage of NYC’s open data sources including their dataset of surface parking, and the PLUTO dataset which includes information about parking garages. With these 2 datasets we were able to quantify the area of parking on every parcel in New York City, as well as the type of parking. To find the number of parking spots, we divided the area of parking by 300 sf, which is a conservative estimate of the amount of area you need per vehicle. This includes both the parking space itself, as well as the drive isles needed to access that space. We found that there are 4 main parking lot types: 1) single use surface parking covering the whole site, 2) accessory surface parking for a building, 3) standalone parking garages, and 4) parking garages built into a larger building.  


The next step was determining how large a building could be built on the site given zoning code and constraints in terms of construct-ability. Again, we used the PLUTO dataset to determine the allowable built density. The measurement of FAR (floor area ratio) specifies the total built area allowed as a ratio of the lot area. For example, a FAR of 1 would allow you to build a one-story building across the entire parcel, or a 4-story building on a quarter of the parcel.  


“We focused on putting the user in control of the potential future scenarios”

Then, we determined the percentage of the parcel that could actually be developed. This is where the parking lot types came in. For the “full lot surface parking” and “standalone parking garages”, the entire parcel can be redeveloped. For “accessory surface parking” only the parking lot can be redeveloped because there are still buildings on the other part of the site. The “parking garages built into a larger building” cannot be easily redeveloped. Many of them have low floor to ceiling heights or are in the interior of buildings and don’t have any daylight access. Therefore, we decided to omit them from the parcels that can be redeveloped. 


Once we determined how much could be built on each parcel, we assigned each parcel a total potential 1) area of park space, 2) number of housing units, and 3) number of workers. The park space was determined by the buildable area of the site, the number of housing units assumed 1000sf per unit, and the number of workers assumed 200sf per worker. For residential and office buildings we removed 20% from the total building area assuming that lobbies, hallways, elevators, and fire stairs would require floor area. With these calculations, we can understand the potential of converting each parking lot to any one of these 3 uses. 

These are just 3 scenarios that can be generated using the slider.

These are just 3 scenarios that can be generated using the slider.

Planning for a Spotless Future

When exploring the data in the web app, we focused on putting the user in control of the potential future scenarios. We did this by creating a simple slider that allocates percentages of parking lots to new uses. The slider picks parcels at random from the entire city. So, if you are making 25% of parcels commercial, the app will select a random 25% of parcels and then sum the total number of workers they could accommodate. 


What we can learn from this is that there are vastly different outcomes depending on what percentage of each use gets developed. Without any specific policy or planning, it’s likely that parking lots will get redeveloped as their new highest and best use as determined by the market. With a tool like this the city can dynamically explore different outcomes and be proactive and introducing planning and policy to achieve their desired results. In addition, this tool makes this data accessible to the public that can then engage the public sector in a more informed way. Too often, planning knowledge is communicated in lengthy reports which prevents it from being widely understood. This type of tool provides a model for how the city can inform and engage its citizens and better serve the public. Check out our other tools here.