Parking-aware navigation system could prevent frustration and emissions
Source: MIT News - AI
Overview
Every day, a motorist checks a navigation app, only to discover that no parking spots are available when they arrive. By the time they finally park and walk to their destination, they’re significantly later than expected.
Most popular navigation systems send drivers to a location without considering the extra time needed to find parking. This not only creates headaches for drivers—it worsens congestion, increases emissions, and can discourage people from taking mass‑transit because they don’t realize it might be faster than driving and parking.
MIT researchers tackled this problem by developing a system that identifies parking lots offering the best balance of proximity to the desired location and likelihood of availability. Their adaptable method points users to the ideal parking area rather than directly to the destination.
In simulated tests with real‑world traffic data from Seattle, the technique achieved time savings of up to 66 % in the most congested settings—roughly a 35‑minute reduction for a motorist compared with waiting for a spot to open in the closest lot.
While a production‑ready system has not yet been built, the demonstrations show the approach’s viability and suggest pathways for implementation.
“This frustration is real and felt by a lot of people, and the bigger issue here is that systematically underestimating these drive times prevents people from making informed choices. It makes it that much harder for people to shift to public transit, bikes, or alternative forms of transportation.”
— Cameron Hickert, MIT graduate student and lead author
Hickert is joined on the paper by Sirui Li, PhD ’25; Zhengbing He, research scientist in the Laboratory for Information and Decision Systems (LIDS); and senior author Cathy Wu, Class of 1954 Career Development Associate Professor in Civil and Environmental Engineering (CEE) and the Institute for Data, Systems, and Society (IDSS) at MIT, and a member of LIDS. The research appears today in IEEE Transactions on Intelligent Transportation Systems (arXiv pre‑print).
Probable parking
To solve the parking problem, the researchers developed a probability‑aware approach that considers:
- All public parking lots near a destination
- The driving distance from the origin to each lot
- The walking distance from each lot to the final destination
- The likelihood of successfully parking in each lot
The approach, based on dynamic programming, works backward from desirable outcomes to compute the optimal route for the user. It also handles the case where a driver reaches the “ideal” lot but cannot find a space. The algorithm evaluates the distances and success probabilities of alternative lots and may recommend a slightly lower‑probability lot that is closer, thereby reducing overall expected travel time.
“If several lots nearby have slightly lower probabilities of success but are very close to each other, it might be smarter to drive there rather than heading to the higher‑probability lot and hoping for an opening. Our framework can account for that.” — Cameron Hickert
The final output is the optimal lot that minimizes the expected total time required to drive, park, and walk to the destination.
Because drivers do not park in isolation, the method also incorporates the actions of other motorists, which affect an individual’s parking‑success probability. Examples include:
- Another driver taking the last spot in the user’s preferred lot
- A driver failing to park in a nearby lot and then taking a spot in the user’s preferred lot
- Spill‑over effects from drivers parking elsewhere that reduce overall availability
“With our framework, we can model all those scenarios in a very clean and principled manner.” — Cameron Hickert
Crowdsourced parking data
Accurate parking‑availability data can come from several sources:
| Source | Description |
|---|---|
| Fixed sensors | Magnetic detectors or gate counters that track cars entering/exiting a lot |
| Crowdsourced reports | Users tap “no parking” (or “spot available”) in an app |
| Vehicle telemetry | Autonomous or connected vehicles report open spots they pass |
| Circling‑vehicle detection | Counting cars that loop around searching for parking |
Since fixed sensors are not widely deployed, the researchers evaluated crowdsourced data as a more feasible alternative. Their simulations showed that crowdsourced observations have an error rate of only ~7 % compared with actual availability, indicating that such data can reliably feed the probability model.
Results
- Congested urban settings (Seattle) – The approach reduced total travel time by ≈ 60 % compared with waiting for a spot to open, and by ≈ 20 % compared with a naïve strategy of continuously driving to the next closest lot.
- Suburban settings – Similar, though slightly smaller, improvements were observed.
- Crowdsourced data accuracy – ~7 % error relative to ground‑truth availability.
Future work
The team plans to:
- Conduct larger‑scale studies using real‑time route information across an entire city.
- Explore additional data sources, such as satellite imagery and computer‑vision detection of open spots.
- Refine the dynamic‑programming model to handle real‑time updates and multi‑user coordination.
Potential emissions reductions
“Transportation systems are so large and complex that they are really hard to change. What we look for, and what we found with this approach, is small changes that can have a big impact to help 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.