ARGUS AI
AI Obstruction Detection
ARGUS AI is computer vision. It reads a roof, detects the obstructions on it and applies the setbacks you specify. Included with both licenses.
Rooftop
Runs on flush mount, east-west and fixed tilt under 20°, from a button inside the roof face.
Scored
A confidence score on each detection, and a sensitivity threshold you set.
Both licenses
Part of the Design License and the Engineering License. No add-on.
How does ARGUS AI work?
Create obstruction templates for different jobs
HVAC setback rules are different across different geographies. With Solesca's obstruction templates you can set these proper setback rules and let AI do the work of finding the obstructions.
- Vents, HVAC, skylights, drains, wires and general obstructions, drawn in 3D on the roof.
- A type on every detection, so the height and setback for that type apply as it lands
- A confidence score on each one, and a sensitivity threshold that keeps or drops the doubtful ones
Your setbacks, applied the moment it lands.
In under 30 seconds ARGUS will find, define, and lay out tens or hundreds of obstructions on the roof.
- Templates apply at the design level, and the designs follow when you update one
- Change the type, height or setback of one obstruction, remove it or duplicate it. A local edit survives a template update
Auto refinement layers to clean up the design
Remove shaded modules, create walkways to equipment, and remove isolated modules.
- Clean up your design in seconds.
The detail, if you want it.
Everything above, answered plainly. Open what applies to you.
What is ARGUS AI?
The computer vision in Solesca. It reads a roof, detects the obstructions on it and applies user-specified setbacks, so that generated layouts reflect what is physically on the roof rather than what a flat image suggests.
The name is the Greek giant with a hundred eyes. It also stands for Automated Recognition and Geometric Understanding System, which is a fair description of the two things it does: recognize an obstruction and give it geometry.
What does ARGUS AI detect?
Obstructions on the roof. The vents, units, skylights, hatches, wires and drains that decide where modules can go, and which are the reason a layout drawn from a flat image does not survive contact with the roof. Detection runs in seconds on the roof face you point it at.
The detection feeds the layout directly. Rather than tracing obstructions by hand and then keeping that tracing in step with every design change, the obstructions are part of the model the layout is generated against.
Whose setbacks does it use?
Yours. Height and setback live in an obstruction template, per obstruction type, that your team owns and applies to the design. ARGUS applies the template. That division is deliberate: setbacks come from code, from an authority having jurisdiction, and from what your engineers have decided they will defend, and none of those are things a vision model should be guessing at.
The default height and setback are zero, so nothing is applied that your team did not set. Update a template and the designs on it follow, except where an obstruction was edited by hand, which keeps the local value.
It also means the output is auditable. When someone asks why a module row stops where it does, the answer is a setback your team set, not an inference nobody can reconstruct.
What is the confidence score for?
ARGUS is sometimes unsure whether a shape is an obstruction, so it scores each detection and lets you set the threshold. A low threshold keeps more, for a first pass over a busy roof; a high one keeps only the detections it is sure of.
Whatever the threshold, each obstruction can still be kept, removed, duplicated or retyped by hand afterwards. The score is there to make review faster, not to replace it.
Which roofs can it run on?
Rooftop designs: flush mount, east-west, or fixed tilt under 20 degrees. The ARGUS button appears inside the roof face on those designs and nowhere else, so a ground-mount site does not get it. Ground-mount screening is a different job and lives in the GIS layers on the Design License.
It sees through one imagery provider, Google Maps, and the detections align to that map. Where the provider has not flown a roof recently, the roof has to be checked against a newer source before anyone relies on the layout.
How accurate is it?
We do not publish a detection rate, because we have not measured one that would hold across roofs, imagery vintages and obstruction types, and a figure that does not hold is worse than none. What the product does instead is show its confidence per obstruction and let you set the bar.
The evaluation that answers this is a roof you have already surveyed. Run ARGUS on it and compare what it found against what your team found.
Is ARGUS AI a chat assistant?
No. There is no conversational interface, and you do not brief it in plain language. Earlier descriptions of ARGUS suggested otherwise and they were wrong. It is a vision model doing one job inside the design workflow.
We would rather tell you that here than in the demo. If what you need is a design assistant you can talk to, ARGUS is not it, and you should discount that part of any evaluation.
Does ARGUS AI choose the setbacks?
No. Setbacks are specified by you, in the obstruction template. ARGUS applies them, which keeps the layout auditable against the code and jurisdiction requirements your team is working to.
Which license includes it?
Both. ARGUS AI obstruction detection is part of the Design License and therefore part of the Engineering License too. It is not an add-on and it does not consume SOL Credits.
Point it at one of your own roofs
Bring a rooftop project you have already surveyed and compare what ARGUS finds against what your team found. Thirty minutes with our U.S.-based engineering team.