ARC turns low-cost commercial UAVs into accountable AI-assisted sensing assets. The drone stays light in the air, the ground station runs the heavy inference, and the operator remains responsible for mission decisions.
ARC lets standard commercial drones become AI-assisted surveillance assets without expensive onboard computers, payload changes, or aircraft redesign.
Onboard AI can reduce endurance, while naive offboard streaming can add latency. ARC bridges that gap by offloading intelligence to the operator workstation while keeping the live view fast.
ARC assists the operator with detection, target focus and mission context, but it does not replace human authority. Alerts remain operator-confirmed and traceable.
ARC can run on a local tactical network, with or without internet. A laptop can host the stream server, run inference, display the mission view and keep data under local control.
Start the local stream server from the ARC window and give the drone controller the exact publish URL it needs. ARC handles ingest, preview and AI-ready output.
The UAV acts primarily as a mobile sensing platform. Heavy AI processing is offloaded to the ground station, reducing onboard computing and power demands.
The AI monitor is built around the freshest available frame. It drops stale frames automatically and offers LIVE or SYNC mode depending on whether speed or exact box alignment matters most.
ARC highlights detected objects with clean tactical boxes, shows all targets on a minimap, and promotes the most relevant target to a PiP close-up for fast operator review.
The AI supports surveillance, but the operator remains in the loop. Force alerts are human-triggered and exported as structured mission records.
ARC can reserve multiple channels, each with its own path, preview link and live status. Operators can see which drone is online and which stream is ready for AI.
ARC is not locked to a single AI model. The operator can open the model inventory, review available trained models, see the classes each one understands, and choose the detector that best fits the mission.
This makes low-cost UAVs more capable without moving intelligence onto the aircraft. Larger or more specialized models can run at the ground station, where power, cooling and compute are easier to manage.
When the AI Agent starts, ARC uses the fastest suitable runtime available. OpenVINO on Intel GPU is preferred when exported models are present, with CPU fallback when needed. The live video stays independent from inference so the operator keeps the freshest view.
ARC — Aerial Reconnaissance and Classification — is a specialized AI-assisted UAV surveillance system designed to extend situational awareness without turning the drone into an autonomous decision-maker.
The architecture keeps the UAV lightweight and shifts intensive inference to the operator workstation. This preserves flight endurance, supports stronger models, and avoids unnecessary onboard payload and power constraints.
Built in Europe, ARC is designed around human supervision, data control and field reliability. It supports responsible deployment by keeping the operator in command while AI accelerates detection, tracking and mission reporting.