Parking lot geometry is a key part of designing a parking facility. It determines how much space a vehicle has to move in order to park. Ideally, a parking lot should be divided into modules based on the angle between an access aisle and a row of parking stalls. An aisle can be anywhere from twelve to twenty-six feet wide. Parking stalls are measured perpendicular to the aisle, so if the angle is less than ninety degrees, the width will be greater than if the stall is 90o off the aisle. However, if the angle is higher than ninety degrees, the front bumper overhang will be reduced to approximately two feet.
The same principle applies to parking lots. A linearly decreasing preference function (ph(x, y)) decreases linearly from the center of the front row to every other location in the parking lot. Using the radial preference function (R), preference decreases linearly. In the case of a radial preference map, preference decreases by an order of two. In the first case, a driver’s preference decreases linearly; the second scenario shows that the preference decreases by an order of two. This is not true in the case of the radial preference function. In a radial preference function, the driver’s choice is radial from the front row to the center of the lot.
The data of parking lots should be combined into a single geometry. This way, AI can “see” parking lots as two-dimensional polygons. It can delineate them using satellite imagery, but it will not be accurate at matching the boundaries of a real parking lot. In this way, the algorithm can only distinguish between parking spaces with a maximum capacity of six cars and parking lots with no more than six cars. The data should be filtered to minimize the influence of a faulty filter.
