What We Do
One platform. One pipeline. A growing set of verticals.
The core capability — fly a defined area, capture systematic imagery, and derive actionable intelligence from it — is domain-agnostic. What changes across verticals is only the AI model trained on top and the report format delivered. RoadGuard, our first product, surveys road corridors and scores pavement condition. BridgeGuard, bridge and culvert inspection, is next on the roadmap — reusing the same core with no re-architecture.
How a survey becomes a report
Open source, by design
We chose PX4, WebODM, and RF-DETR deliberately, not by default. Building on mature open-source foundations — instead of costly proprietary flight, photogrammetry, and detection software — keeps our own cost base down, and that saving flows directly into more economical, outcome-priced surveys for our clients. It also puts RoadGuard in step with the Government of India's own stated direction: MeitY's Policy on Adoption of Open Source Software for Government of India calls for open source to be the preferred choice over closed systems in e-governance wherever it's viable.
- PX4 / MAVSDK
- Autonomous flight software for survey missions
- WebODM
- Photogrammetric reconstruction into an orthomosaic and elevation model
- RF-DETR
- The detection backbone, chosen to keep the pipeline commercially unencumbered
It's also a Make-in-India story in the truest sense — not importing a foreign black box, but assembling and extending open building blocks into a complete, Indian-built survey and reporting pipeline. Our own engineering sits on top of that foundation: the fine-tuned detection models and the India-specific, IRC:82-labelled dataset we build with every survey flown, refined here rather than published. As RoadGuard proves itself in the field, it's also a proof point for the open-source projects that made it possible — the kind of success story that helps grow the ecosystem it was built on.
Where we are: the detect-to-report pipeline runs end-to-end today. Field validation against real, geo-tagged Indian road imagery — fine-tuning the detector on data collected during pilot surveys — is the current, active focus.