Adaptive fixed + mobile sensing
Fixed cameras, drones and mobile units combine into adaptive coverage that follows the work.
EPFL Spinoff · Lausanne, Switzerland
Raven turns cameras, drones and AI into actionable site evidence across safety, security, materials and 3D site status, helping teams act before risks escalate into accidents, thefts, delays or disputes.
Backed by
The product
Construction sites still rely on fragmented visibility: manual inspections, fixed cameras, drone footage and disconnected reports. Raven turns this into decision-ready evidence across safety, security, materials and 3D site status.
Built privacy-by-design: face blurring, controlled data retention and audit-ready evidence workflows.
Findings arrive as clear, time-stamped evidence teams can act on, not raw footage to sift through.
One pipeline across safety, security, materials and 3D site status, instead of disconnected point tools.
Surface risks early, so issues are resolved before they become accidents, thefts, delays or disputes.
Reports generated automatically, ready for stakeholders and compliance.
How it works
Five integrated stages, from raw capture to audit-ready evidence.
Fixed cameras, drones and mobile units capture the site continuously, indoors and out.
AI reads each scene across 2D and 3D, detecting risks and changes in near real time.
Findings are cross-checked across sensors and time to remove noise and false positives.
Verified risks reach the right teams immediately, so they can act before issues escalate.
Outcomes are logged as audit-ready evidence, closing the loop on every finding.
Technology
Raven brings sensing, indoor and outdoor adaptation, validation, scene intelligence, privacy and reporting together into one pipeline.
Fixed cameras, drones and mobile units combine into adaptive coverage that follows the work.
Robots and models adjust to lighting, GPS-denied interiors and open-air sites, so coverage holds across every environment.
Signals are cross-checked across sensors and time, turning raw data into evidence you can trust.
Detection, segmentation and 3D reconstruction read the site across imagery and reconstructed space.
Face blurring and controlled retention keep people protected and identifiable data out of the model.
Decision-ready reports package findings into audit-ready evidence for every stakeholder.
Use cases
One adaptive pipeline, four kinds of decision-ready evidence.
PPE gaps, danger-zone breaches, unprotected edges and machinery proximity, captured as visual proof.
Intrusions and asset movement flagged as time-stamped evidence for security teams and insurers.
Track deliveries, stockpiles and material flow, so teams know what is on site and where.
Reconstructed 3D site state to monitor progress and settle delays or disputes with facts.
Unlike fixed cameras, safety-only AI alerts or passive drone capture, Raven combines adaptive coverage with decision-ready evidence across multiple site domains.
Performance
Faster than manual site inspection workflows.
Cost reduction relative to traditional inspection approaches.
Continuous monitoring, with no shifts and no downtime.
Where we are
Raven Intelligent Systems is an EPFL-rooted startup building robotic site intelligence for high-risk worksites. Our MVP-ready system combines fixed sensing, drone-assisted capture and AI-supported evidence generation.
At the EPFL Double Deck construction site, we are running an industry-shaped pilot with field deployment, structured reporting and feedback from construction and safety stakeholders.
Become a pilot partnerPilot-ready MVP. Pre-seed stage. EPFL Double Deck pilot running.
Paid-pilot deployments with full five-stage pipeline.
Swarm-scale deployments across construction, energy and manufacturing with expanded hazard coverage.
Get in touch
Whether you are an owner, construction or QHSE/security leader, operations or innovation lead, project/integration consultant, or an investor interested in Raven's next stage, we would be glad to connect.
Built privacy-by-design: face blurring, controlled data retention and audit-ready evidence workflows.