The 4 AI Engines of ROQUE.AI

---

The Optical Object Tracker is an Image Recognition Engine responsible for the integration of visual data from the site.

The Georeferenced Status Visualisation Engine is the backbone of the project's digital twin we call VisualTwin.

The Future State Forecaster is a Schedule Simulation Engine which allows for Diagnosis and Cure of overruns.

And finally, the Craftsmanship Dual Intelligence is a deep-learning Engine which accumulates performance data from diverse projects. 

ROQUE.AI utilises state-of-the-art image recognition algorithms to identify presence of components and progress of works, analysing footage of any type (satellite, drone, 360 etc.).

For open-space projects, satellite imagery could be readily available upon request, when 50 cm precision is adequate.

Drone imagery could be used in case of long periods of cloudy weather or when higher accuracy is necessary (cm precision).

Visual data complement site data adding an extra layer of certainty and proof.

VisualTwin is the proprietary digital twin platform of ROQUE.AI. It operates a tempo-spatial “map” of the actual current, past and future project status.

It consolidates, "georeferences" and archives all construction related project information (photos, tickets, serial numbers, drawings etc.) valuable also during the operational phase.

Predominantly, it provides to all stakeholders an eagle's view on the site, "live" detailed reporting on construction progress, and automatic detection of delays.

ROQUE.AI utilises state-of-the-art algorithms to simulate multiple “what-if” construction scenarios in a fraction of the time and cost of traditional methods.

Through Diagnosis, the expected overrun is quantified, whenever delays are detected by VisualTwin or “unforeseen” events occur (delayed deliveries, bad weather, shortage of resources etc.).

To Cure the overruns, optioneering on potential mitigation measures reveals the optimum blend, in line with project priorities (cost vs time sensitivity)

ROQUE.AI utilises its proprietary deep-learning algorithm, trained continuously on data of diverse projects, to check the feasibility of the schedule and to process progress data of VisualTwin.

Replacement of schedule’s assumptions with a blend of historic and site data improves materially the accuracy of the Diagnosis and Cure results.

Measurable performance targets are set and resources’ adjustments are exactly quantified, with their effectiveness assessed continuously by VisualTwin.