Counting Vending Machine Foot Traffic Where Cameras Can't Go: Millimeter Wave Radar

Your vending sales report has a blind spot.

It tells you what sold.

It does not tell you who walked past and didn’t buy.

That number is foot traffic. Without it, a poorly performing machine is a guess.

Two problems that look the same in a sales report

Say a machine sells ten items a day. That can mean two very different things.

Case 1 — low traffic. Forty people pass the machine each day and ten buy, a 25% conversion rate. The machine is working. The location isn’t. Move it.

Case 2 — low conversion. Eight hundred people pass each day and ten buy, a conversion rate of about 1%. The location is excellent and the offer is failing. Change the products, the price, the payment options or the screen before you think about moving it.

The sales figure is identical, but the right fix is the opposite. (These figures are illustrative, not from a deployment.)

Only a traffic count tells the two cases apart.

The obvious sensor is a camera, and that’s the problem

The standard way to count people is a camera with analytics. It works, and where a camera is appropriate it can even tell you whether someone looked at the screen.

But a camera captures images of people, and 2026 made that a procurement question, not a footnote. Australia’s privacy regulator now asks for a “precautionary approach” to facial recognition in retail. We covered what that means for vending cameras in Australia just drew a stricter line on facial recognition.

And in some places, the question is already answered. You don’t put a camera in a restroom. Or a change room, or most clinic corridors.

Those are exactly the places where many small machines live: wall-mounted dispensers for sanitary products, hygiene items, sexual health kits and test kits.

The machines that most need a traffic count are often the ones where a camera cannot count.

Radar counts people without seeing them

A millimeter wave radar sends a low-power radio signal and reads the reflections.

From those reflections it calculates where each person is, which way they’re moving and how fast. Software tracks those targets and counts them in and out of a zone.

It never captures an image. There is no face to recognise and no footage to leak. It works in the dark and it isn’t affected by lighting.

This is not a theoretical idea. We built it and filmed the demo:

In the demo:

• The radar is mounted at 1.9 m, about the height of a KioskForce beverage vending machine.
• It connects over USB to the traffic monitoring software.
• The software counts entries, exits and the number of people currently in the zone, and keeps a running duration.
• As our engineer walks in and out of the zone, the counts update in real time.

The full walkthrough is on the mmWave radar traffic counting feature page.

What operators can do with a traffic count

Diagnose weak machines. Traffic and sales together give you conversion rate, and conversion rate tells you whether to move the machine or change what it offers.

Pick sites before committing a machine. Mount a radar at two or three candidate locations for a few weeks. The one with the most traffic wins, and you have data instead of a hunch.

Find peak hours. Plan restocking and service visits around when people are actually there.

Sell the screen. A machine with an advertising screen is a media placement. An advertiser wants to know how many people were in front of it. A radar count gives that number with no camera involved.

Report on public programmes. Free sanitary product and public health dispensing programmes are often judged on reach. Traffic next to dispense counts shows reach and uptake separately.

Radar vs camera: pick the sensor for the question

Question you need answered	Radar	Camera
How many people passed?	Yes	Yes
How many stopped at the machine?	Yes	Yes
When is it busy?	Yes	Yes
Did they look at the screen?	No	Yes
Can it go in a restroom or change room?	Yes	Generally no
Does it capture images of people?	No	Yes

If the question is “how many”, radar answers it with less to justify. If the question is “did they look”, you need a camera, and a place where a camera belongs.

What to plan for

Radar is not a sticker you put on a cabinet. Each project needs:

• A mounting position. Height and angle depend on the cabinet and the space in front of it.
• A detection zone. Count the area in front of the machine, not the whole corridor.
• A test on site. Metal surfaces and crowds affect radar reflections, so every site gets tested when the radar goes in.
• The right radio module for the market. mmWave radio rules differ by country.
• A reporting path. In the demo, the radar connects to a laptop. In a custom build, it connects to the machine’s controller, so traffic sits next to sales from the same machine.

Where this stands

Radar traffic counting is a technical preview at KioskForce. It is built, demonstrated and ready to go into a custom project.

Because every KioskForce machine is designed to specification, we can add it to a wall-mount vending machine, a kiosk or a smart locker bank: mounted, zoned and reporting.

If your machines sit where a camera can’t, and you want to know how many people pass them, tell us about the site.