Spark: A Smart Parking Lot Monitoring System

Abstract

Parking on a college campus is understood to be a challenge for commuters. With a rising matriculation rate in the United States, the task of finding parking on an expansive campus grows even more daunting. However, the rising prominence of the Internet of Things has initiated a paradigm shift in data-analysis computing. The point of data collection is often outlier locations, removed from existing infrastructure, and parking lots are no exception. Using proximity sensors, solar power, and cellular communication, we can create such an IoT system to monitor parking lot in- and outflows. The parking data collected can be analyzed to create a smarter, more efficient parking experience.

A Sensor Calibration

The VCNL4200 sensor does not provide a proximity reading in distance units. Rather, it provides a unitless, 16-bit figure. We need to be able to attach a distance meaning to this figure in order to determine whether or not the reading corresponds to the height of a car’s chassis.

Fig. 7.

Proximity data vs. distance (in.) for VCNL4200.

In order to effectively configure our sensor to detect a car, we need enough current through the IRED to overcome the lack of reflectivity of a car’s underside but not so much that we drain our energy reserves. We used a piece of plain cardboard with medium reflectivity, placed at successive 2 in. intervals along a tape measure, to construct characteristic curves for the VCNL4200 under our operating conditions (1/320 duty cycle, 250 measurements/s, 200 mA IRED, 16-bit output). These conditions can be changed to reduce energy consumption or improve range.

The curves recorded in Fig. 7 are clearly exponential but also display quasi-linear behavior beyond a certain point. We moniker this value the “Point of Consistency,” because results measured at and after this distance are repeatable regardless of precision. At closer distances, two to four inches especially, the output values vary too widely to be determinant. At around 14 in., however, the results were consistent within ±2 (for the resistances shown above). Our target measurement height is 8 in., but designing for a factor of safety of 2 requires we target 16 in. The 68 $\Omega$ resistor displays linear behavior at 16 inches and beyond, thus we select that resistance value.