A near real-time approach to sense parking across the city
An iOT platform, Tessel is used to build a light sensor that transmits light readings every 1/2 a second over SMS (text messages). The approach is used to build a movable parking sensor platform to communicate parking information in near real-time
This project was inspired by deliberate thoughts on how to improve the parking experience by finding ways to communicate parking awareness to drivers. An IBM survey found that, on average, drivers globally spend 20 minutes searching for parking. Thats a lot of needless emissions and waste.
Current approaches suffer from a lack of precision and timeliness in order to inform a driver of the exact location of a parking space within a certain acceptable time-period. To overcome this in a brute-force manner, existing approaches include smart parking meters where every parking spot has to be equipped with a fixed sensor which is costly and doesn't scale well.Im also not entirely convinced that people sharing information about open parking spots is a reliable solution.
The need here is to effectively measure and transmit occupancy of an entire street segment passively and do it in a manner that can be scaled to thousands of street segments across the city.
Our approach is to demonstrate the notion of a movable sensor platform to communicate the same information with a slight trade off in time but maintain precise location information
The basic idea is to measure light readings from underneath parked cars on a typical NYC street. The sensor itself can be mounted on top of something that can moves along a programmable path.
- The light sensor is used to collect light readings every 500 ms.
- <Light reading, Location,timestamp> are transmitted over SMS every <x> seconds to a server.
- A moving client (Car with a navigation app) queries the server repeatedly using<timestamp, location> to get a response that contains real-time (current to about 2 minutes) parking availability.
- A single street segment can have 2 sensors on either side running up and down the street. Check out the NYC LION Street File for a digital catalog of all the city's street segments.
This approach assumes that the sensor is able to move along the side of a street of a common residential street. The engineering challenge is to be able to install it in a manner that takes into account the elements and the fact that vehicles will likely run over this thing . Looking forward to interesting ideas in getting around this. The other challenge is baking in parking restriction data like the location of fire hydrants etc.
We have not yet found a dataset that contains the locations of fire hydrants although a zanyidea to do this programmatically was explored here
PS: A friend and I also discussed the idea of using Beacons on street corners that simply record the entry/exit of vehicles and thereby derive an aggregate parking awareness. It is an interesting approach as well that deserves further study.