Context & Challenge
Parking lot attendants cross the city daily to check if all cars are correctly parked, or if the driver has paid. They used to do this with a predetermined walking (or cycling) route, with a total of 53 fixed routes. Each attendant was provided with a paper map and directions to follow. These walking routes were generated based on a very limited number of parameters. If they wanted to adapt these walking routes, they had to go through a complex system which took a lot of time; multiple days to be exact. There were several drawbacks and inefficiencies, such as:
- Some streets were not regularly checked while others appeared in various routes.
- The parking lot attendants did not have any autonomy or flexibility to change their route, e.g., in case of road works or street events which made parking impossible
After conducting a pilot project whereby routes where manually recalculated based on a broad set of (additional) parameters, the city decided to go for a brand-new application that would allow to generate new routes in minutes, compare them and finally publish them to be used the next day. Besides a set of parameters, the algorithm also has to consider additional limitations as well as datasets containing GIS data.
The business challenge in short: check as many parked cars as possible. Easier said than done.
The project challenges included:
- Delivering a system that can be used by businesspeople to manage all parameters and to generate optimised routes.
- Delivering the algorithm that calculates these routes.
- Delivering output necessary to publish the routes for further use.
Pencil42 proposed to add yet another component: an app to be used by the parking lot attendants. Every morning, this app provides them with all the necessary information to perform their jobs properly, allowing them to prioritise within a route and track the streets that were visited during the day. This allows for next-day optimisation of routes and setting priorities based on specific road conditions that were encountered during the day.
Our answer to all their questions
PENCIL42 started by identifying:
- Business Process Models
- Information Models
- User Stories
- Wireframes
- A first version of the complex optimisation algorithm
This resulted in a functional architecture. Following the scrum approach, we incrementally delivered the application, considering feedback from the different end-users. This resulted in:
- An application that allows you to manage and change the different parameters on the fly.
- Ease of use, which was crucial and considered from the very beginning.
- A solid (complex) algorithm that considers all parameters, constraints, and datasets.
- Fast and accurate generation of walking routes.
- Publication of optimal routes.
- An app for the parking lot attendants that gives them flexibility, autonomy, responsibility, and ease of reporting anomalies detected during their tour. This information is then considered for regenerating these tours. The app also helps the parking lot attendant to see which streets still have to be checked.
- Running and monitoring the application in the cloud
Today the city of Ghent has improved its efficiency in managing parking lot spaces resulting in a higher satisfaction rate of the parking lot attendants.
Technology used
- Google OR-Tools
- Solver
- ….
