Online — in real time
Live detection as input for adaptive control.
- Detection and classification
- Queue length per approach
- Speed and density
Works with PTV Epics and PTV Balance.
SMART LIGHT
HAWKAI — AI Video Traffic Analytics
A loop sees a point. A camera sees the intersection.
HAWKAI processes live video for real-time traffic control, or recorded video for detailed analysis and strategic planning. It works with fixed cameras on a pole or with a drone, and detects, classifies and tracks every road user in the frame. The output: origin–destination matrices, detections and trajectories, ready for control and planning.
What HAWKAI measures
- Origin–destination matrix
- Vehicle detection
- Vehicle classification
- Vehicle speeds
- Density estimate
- Queue lengths
- Delay (time losses)
- Pedestrian detection
- Trajectory and movement analysis
Offline — recorded video
Four cameras merged into one trajectory map for analysis and planning.
- O–D matrix per manoeuvre
- Trajectories and occupancy heat map
- Delay and demand per lane
Works with PTV Visum, PTV Vissim and LISA.
Loop vs camera
Inductive loop
Sees only the vehicle inside the loop's zone.
HAWKAI video detector
Sees the whole approach: every vehicle, the queue and the speed.
What you get from a recording
Origin–destination matrix
Vehicles for each entry–exit pair.
Vehicle trajectories
Every path through the intersection, coloured by manoeuvre.
Occupancy heat map
Where traffic actually stands still: queues and conflict zones.
Example output
| From \ To | A | B | C | Σ |
|---|---|---|---|---|
| A | · | 54 | 220 | 274 |
| B | 92 | · | 21 | 113 |
| C | 348 | 21 | · | 369 |
| Σ | 440 | 75 | 241 | 756 |
