Compared to per-space sensor solutions, 2-5 car counter sensors in a car counting mode sometimes can replace around 200-900 per-spaces sensors.
But, as always, there are some trade-offs!
Cost vs. Accuracy
There are a few fundamental factors that make car counting systems inherently less accurate than per-space sensors.
The per space sensor has all the time to detect the car. Once a vehicle has parked, it can also verify its detection with multiple repeated measurements and then check the vehicle’s presence every few seconds. The car counter often has just a couple of seconds to detect a passing vehicle. And sometimes it can be one long vehicle and sometimes – two short tailgating cars.
No detection is 100% accurate, a per space sensor can also miss a car but there is an important difference in the error treatment mechanics between per space and car counter sensors. In every car counting system (including induction loops, wired solutions, etc.) an absolute occupancy error is proportional to the total turnover (number of cars counted) and this error builds up over time, the so-called “error accumulation” or “error creep” phenomenon. In the case of per space sensors, since each sensor resets itself every time a parking event takes place – it is always a direct measure of the number of spaces/sensors in the lot. Let’s consider an example where we have a 99.5% accurate system of two types installed in the lot of 100 spaces and a daily turnover of 100 cars. In one month, we will have a 3,000 vehicles turnover and a 0.5% error gets us an expected 15 spaces / 15% error for a 100 spaces lot. And in two months the error will double. In most cases, it is only a question of time when the counting error will creep to an unacceptable level. Per space system with no error accumulation, on the other hand, will remain, 99.5% accurate. In the same case of a 100-space facility, it will always remain 99.5% accurate, meaning a maximum error of 0.5 vehicles/spaces at any given time.
When does the car counting system work well?
Based on our experience a combination of two factors makes a good case for car counters as a lower cost and acceptable accuracy alternative for per space sensors.
Well Defined One-Way Entry and Exit Lanes
Broad bi-directional entries and exits are more prone to counting errors. Even a little accuracy reduction from 99.5% to 98%, that may take place in such an environment, can result in an unacceptable 15%+ error in just one week of operation. Narrow, well-defined, one-way, straight entry and exit are much easy to operate with high counting accuracy which can maintain a reasonable total availability estimate within a week even in the case of a high turnover lot.
Below are two examples of “bad” and “good” counting configurations. On the left – an example of data flow for a counting solution that has an inherent accuracy deficiency, leading to error accumulation over time. Such deficiency could be due to limitations of a counting hardware, or, in case of using a ground-based sensor, due to broad entries/exits where a vehicle may missing a counting sensor while entering or exiting the facility. On the right – an example of a “good” counting solution deployment with no error accumulation.