Traffic counting methods compared: tube, radar, camera or LiDAR?
Every method counts vehicles. They differ in what else they see, how long they last, what they cost to move, and what they mean for privacy in a public street.
On this page
Traffic counting methods are the ways to count vehicles by direction and hour on a road or street. This guide compares pneumatic tubes, radar, cameras and LiDAR on what each records, where each fails and what each means for privacy. Bumbee Labs is the company behind the only footfall method in Europe approved by a data protection authority.
How do tubes, radar, cameras and LiDAR compare for traffic counting?
| Method | What it records | Privacy | Installation | Best suited to |
|---|---|---|---|---|
| Pneumatic tube counter | Axle pulses from rubber tubes across the lane | Registers axle pulses | Tubes across the lane that wear and detach | A short count on a free-flowing road |
| Radar traffic counter | Vehicle presence and speed; a narrower picture in mixed urban movement | Captures nothing resembling identity | Permanent, weather-exposed, little maintenance | A permanent station on open road, such as a motorway |
| AI traffic counting camera | Counts inside the camera and outputs numbers | No personal images stored in counting mode, but perceived as surveillance on a public street | Reuses camera hardware many organisations already run; night and glare performance depends on the light | Site entrances and service points |
| LiDAR | Moving geometry from eye-safe laser pulses (position, size, direction and speed of movement); classes separated | Never creates an image | One installation counts continuously in any light; a newer technology, and not the cheapest per point | A street with mixed movement, continuous counts or a privacy-sensitive public setting |
Which traffic counting methods fit a road or a street?
Tube, radar, camera or LiDAR: which one suits our street?
A tube across the lane suits a short count on a free-flowing road, and radar suits a permanent station on open road. For a street with mixed movement, continuous counts or a privacy-sensitive public setting, laser traffic counting with LiDAR fits. An AI camera suits site entrances and service points.
How do we know the count is right?
The count is checked against a manual control count of the same street over the same hours. Bumbee Labs verifies each deployment this way, matching manual control measurements to the system’s own data. Any supplier should be able to describe an equivalent procedure; see procuring traffic counting.
Automatic traffic count (ATC) or manual classified count (MCC)?
An automatic traffic count (ATC) puts a counter on the road, such as a tube; a manual classified count (MCC) puts a person at the roadside. Both are often short samples. From a continuous series you can read the average for any period, as with AADT.
Why does a tube count go wrong in queues and at low speed?
A pneumatic tube counter registers axle pulses from rubber tubes across the lane, and its accuracy drops when traffic queues or crawls. Each count is a sample of a week or two. Tubes also wear and detach, and they cannot see pavements or cycle paths (see bicycle counting).
Does counting still work in the dark, in glare and in snow?
That depends on the sensor. AI camera counting is still affected by the light at night and in glare, while radar copes with weather and darkness and LiDAR measures in any light. Outdoors on a street, the housing must handle rain, snow and direct sun, so each deployment is configured for its site.
When do fixed count points pay off against hired tube counts every year?
A tube campaign is inexpensive, but every repeat costs money again and produces a new sample. LiDAR hardware is not the cheapest per point. Counted per question answered, one continuous series from a street usually gives more for the money than a stack of single-purpose samples, and the same series, delivered as data deliverables, can be reused for later questions.
Which other questions can traffic counting methods answer?
- Radar on the approach road, LiDAR in the centre: can the methods be mixed? Yes, methods can be combined; we design traffic counting per site with LiDAR and AI cameras, and that data goes into the same platform as pedestrian and bicycle data; see hybrid solutions.
- Traffic between towns and districts, not at one point: what measures it? Catchment analysis from mobile data shows which districts people come from, per day and season, and complements the counts at a point.
- Can our short counts be scaled up to AADT with a few fixed points? Scaling up a short count rests on assumptions about the other months; a continuous series from a traffic survey gives AADT directly.
Curious what this looks like for your venue?
Frequently asked questions
What is the most common traffic counting method for short-term road counts?
Pneumatic tubes remain the workhorse for short-term road counts: two rubber tubes across the lane, counting axle pulses for a week or two. They are cheap and proven, but each count is a short sample rather than a continuous series. Tubes also wear out, struggle with slow or queueing traffic, and say nothing about pedestrians or cyclists on the same street.
When is radar the right choice for traffic counting on open roads?
Radar suits permanent, weather-exposed installations focused on vehicle presence and speed on flowing roads. It is robust and privacy-neutral, but gives a narrower picture in complex urban environments with mixed movement, queues and crossing flows.
What does LiDAR add over cameras for traffic counting on a city street?
Both see complex scenes, but LiDAR measures geometry with laser light and never produces an image, so there is no footage to store, mask or justify in a public street. It also works identically in darkness and glare. Cameras with built-in counting output numbers rather than footage in counting mode, but the optical hardware itself carries a surveillance perception that matters in public space.
Can one traffic counting system count vehicles, cyclists and pedestrians together on a street?
Yes. This is where modern sensors do better than single-purpose equipment. A LiDAR deployment can be configured to separate movement classes, so one street reports vehicles, cyclists and pedestrians as distinct series in one dashboard, which is what mixed-use urban planning questions need.
Where this fits
Part of
Industries and segments
Challenges
Metrics
Platform and integrations
Data protection:GDPR-compliant footfall analyticsThe IMY decisionGDPR & traffic measurementEU AI Act