Solutions by typeMobility analysis
How long has that been lying in the lane?
Something is in the lane and nobody has seen it
Twenty past eleven in the morning, inside a tunnel. A lorry has shed part of its load on the bend: two burst boxes, some tubes and a pallet lying across take up the right-hand lane. The lorry has driven on and knows nothing about it. The cars behind swerve into the next lane as they come, one after another, and whoever arrives in forty seconds will not see it until they come out of the bend.
3 cameras across 2 sectors
This is how you ask for it
«Alert me if something falls onto the carriageway, and tell me which lane it landed in.»
An object on the carriageway does not move, does not flee and does not own up. It stays there until somebody sees it and reports it, and that somebody is usually the driver who has already braked hard. Inside a tunnel the margin is even smaller: no hard shoulder, no visibility, and a chain of hard braking is what turns a pallet into a multi-vehicle crash and a tunnel closed for half a morning.
You decide once
You decide this once. You say which stretches are watched, what lanes each one has, what counts as an obstacle and who is told: the control centre, the maintenance patrol, or both at once. IRIS applies it on every camera along those stretches, day and night, and it does not depend on anyone having that quadrant open. If a new stretch opens, you add it to the rule.
This is the opposite of almost everything else a camera does. There is no movement to detect here: there is something sitting still where nothing should be. A pallet in the middle lane, rubble covering the left one, a loose wheel that has rolled twenty metres further on. None of it moves, none of it draws the eye on a screen, and yet every second it stays there is one more car arriving at a hundred and twenty without knowing.
The value is in the lane, not the object. That there is something on the road eventually gets known. What changes the response is knowing which lane it is in and at exactly what point, because that determines whether one lane is closed or two, which sign lights up, which kilometre the patrol is sent to and whether tunnel traffic has to be stopped. That detail is what gets lost today in a driver’s call saying «there is something lying around near the toll».
And then there is the part only the video holds: where it came from. With the recording you know which vehicle shed it and at what time, and that stops being an argument about who pays for the clean-up. The same camera that flags the obstacle keeps the minute it fell, which is the only evidence that exists for something that left no other trace.
What changes
Before, the control centre found out by phone, and whoever called was driving and could not name the lane or the exact kilometre. A patrol was sent out to find it. Now the alert arrives with the camera, the point and the lane occupied, so the sign lights up sooner and the patrol sets off knowing what it is going to and what it will find.
Where we have seen it
CAM-13Load spilled on the road
«Tell me if a vehicle drops its load on the road, and which lane it is blocking.»
CAM-64Load spilled across a lane
«Tell me if something falls onto the road, and which lane it has landed in.»
CAM-35Debris inside the tunnel
«Tell me if an obstacle appears on the tunnel roadway, and which lane it is in.»
What it does not do
IRIS does not remove the load or close the lane: it alerts, says where it is and opens the image, and from there a person acts. It also does not know what the object is: it tells that something is on the carriageway that should not be, not whether it is a pallet or a cardboard box. And in a dimly lit tunnel, or on wet road full of reflections, a low dark object is hard for anyone to see.
Questions
- Does it detect any object, or only large ones?
- It depends on its size in the image, and that is the honest answer. A pallet lying across a lane shows up without trouble; a small part two hundred metres from the camera may fall below what the image resolves. What we do at commissioning is check camera by camera how far each one reaches and above what size it alerts. We would rather say «beyond this point we do not guarantee it» than promise everything is seen, because the second falls apart on the first trial.
- Does it say which lane it ended up in?
- Yes, and it is the most used piece of information. Lanes are marked once on each camera’s image, so the alert does not say «there is something on the road»: it says which lane it is in and at what point. Everything downstream depends on that, from which sign lights up to whether a lane is closed or the tunnel is stopped. Without the lane, the alert forces someone to watch the video and work it out, and those are minutes that count on a fast road.
- Can we find out which vehicle it fell from?
- Yes, if it went past the camera. Because the alert is kept with its exact time, you can go to the video of the preceding seconds and see the vehicle that shed it. That turns an argument about who pays for the removal into a verifiable fact. What we do not do is promise the plate under any conditions: that depends on the camera, the speed, the light and the angle, and it is checked stretch by stretch before we say yes.
- And at night, or inside a lit tunnel?
- It works, with caveats worth knowing beforehand. A tunnel has constant lighting, which helps, but also reflections on wet road, abrupt light changes at the portals and long vehicle shadows. A low, dark object in those conditions is hard for any system and for a driver too. That is why the tuning is done with the tunnel’s real cameras and not from a catalogue, and if a particular stretch cannot be guaranteed, we say so and the framing is reinforced.
The real interface, step by step
A camera looks. IRIS understands what it sees.
An extract of the real screen, no voice-over and hands-free. Each round poses another situation. What sits behind it is a good deal bigger than what you see.
- It lands in the tray
- You open it
- You confirm or dismiss
Not seeing yours here?
These words are the ones we have built, not the ones IRIS understands. The list falls short on purpose: you ask for it in a sentence, and sentences do not run out. Tell us what is in front of your cameras and we will tell you whether it understands it, or not yet.
And the best part
IRIS only sees what the camera shows.Where we do not reach, we say so.
No software fixes a lens pointing the wrong way or covered in dust. Before promising anything, we say which cameras are in a state to see what you are asking of them.
Tell us your problem and we will say whether IRIS understands it, or not yet.
We reply the same working day.