This happens every day – a driver driving through the city checks in the navigation application how long the journey will take, but when he arrives he finds no available parking spaces. When they finally park and walk to their destination, it's much later than they expected.
The most popular navigation systems send the driver to their destination without taking into account the additional time needed to find parking. It's not just a headache for drivers. It can worsen congestion and increase emissions by forcing drivers to wander around looking for a parking space. This underestimation can also discourage people from using public transport because they don't realize it can be faster than driving and parking.
MIT researchers have addressed this problem by developing a system that can be used to identify parking lots that provide the best balance between proximity to a selected location and the likelihood of parking availability. Their flexible method shows users their ideal parking area, rather than their destination.
In simulated tests with real Seattle traffic data, this technique resulted in time savings of up to 66 percent in the most congested areas. For motorists, this would shorten the travel time by approximately 35 minutes compared to waiting for a space to become available at the nearest parking lot.
Although they have not yet designed a real-world-ready system, their demonstrations demonstrate the feasibility of this approach and indicate how it can be implemented.
“This frustration is real and felt by many people, but the bigger problem is that systematic underestimation of commute times prevents people from making informed choices. This makes it much harder for people to switch to public transit, biking or alternative forms of transportation,” says Cameron Hickert, an MIT graduate student and lead author of the paper describing the work.
Hickert is joined in the article by Sirui Li PhD '25; Zhengbing He, research associate at the Laboratory of Information and Decision Systems (LIDS); and senior author Cathy Wu, 1954 Associate Professor of Career Development in Civil and Environmental Engineering (CEE) and the Institute for Data, Systems and Society (IDSS) at MIT and a member of LIDS. Tests is published today in Intelligent Transport Systems Transactions.
Possible parking lot
To solve the parking problem, researchers developed a probability-based approach that takes into account all possible public parking lots near the destination, the driving distance from the starting point, the walking distance from each parking lot to the destination, and the probability of parking success.
The dynamic programming approach works backwards from good results to calculate the best route for the user.
Their method also takes into account the case when a user arrives at the perfect parking lot but cannot find a free space. It takes into account the distance to other parking lots and the probability of successfully parking in each of them.
“If there are several plots nearby that have a slightly lower probability of success, but are very close to each other, it may make more sense to go there rather than go to the plot with a higher probability and hope to find an opening. Our framework can take this into account,” says Hickert.
Ultimately, their system can identify the optimal lot that has the shortest expected time needed to drive, park and reach the destination.
But no driver expects to be the only one trying to park in a busy city center. Therefore, this method also takes into account the actions of other drivers, which affects the user's probability of parking success.
For example, another driver may be the first to arrive at your ideal parking lot and occupy the last parking space. Another driver can also try to park in a different parking lot, but fail to park in the user's ideal parking lot. Additionally, another driver may park in a different parking lot and cause side effects that reduce your chances of success.
“With our framework, we show how you can model all of these scenarios in a very transparent and rules-based way,” says Hickert.
Crowdsourced parking data
Data on parking availability may come from several sources. For example, some parking lots have magnetic sensors or gates that track the number of cars entering and leaving.
However, such sensors are not widely used, so to increase the real-world feasibility of their system, researchers instead investigated the effectiveness of using crowdsourced data.
For example, users can pinpoint available parking via the app. Data can also be collected by tracking the number of vehicles circling looking for a parking space or the number of entries and exits into a parking lot after failure.
One day, autonomous vehicles will even be able to report on available parking spaces as they pass by.
“Right now, a lot of this information goes nowhere. But if we could capture it, even by clicking 'no parking' in the app, it could be an important source of information that allows people to make more informed decisions,” adds Hickert.
The researchers evaluated their system using real-world traffic data in the Seattle area, simulating different times of day in congested urban and suburban environments. In crowded areas, their approach reduced total travel time by about 60 percent compared to sitting and waiting for a space to become available, and by about 20 percent compared to a strategy of constantly commuting to the nearest parking lot with a locker.
They also found that crowdsourced parking availability observations would only have an error rate of about 7% compared to actual parking availability. This means it can be an effective way of collecting parking probability data.
In the future, researchers want to conduct larger studies using real-time route information across the city. They also want to explore additional options for collecting parking availability data, such as the use of satellite imagery, and assess potential emissions reductions.
“Transportation systems are so large and complex that it's really hard to change them. What we're looking for, and what we've found with this approach, are small changes that can have a big impact by helping people make better choices, reduce congestion and reduce emissions,” says Wu.
This research was supported in part by Cintra, the MIT Energy Initiative, and the National Science Foundation.
