SPARC AI has partnered with Rate Manufacturing to integrate its Overwatch navigation software into the Model-F multi-mission drone.
Australia-based SPARC AI has partnered with US firm Rate Manufacturing to integrate its Overwatch navigation software into the Model-F multi-mission drone, targeting demand for systems that can operate when GPS is jammed or unavailable.
The partnership combines Rate’s scalable drone manufacturing platform with SPARC AI’s software.
Instead of GPS, the Overwatch system uses onboard inertial sensors combined with AI-based software to reduce navigation drift and maintain targeting accuracy without requiring additional external hardware.
Rate’s Model-F is a modular drone platform with a top speed above 70 miles (113 kilometers) per hour, an operational range exceeding 6 miles (10 kilometers), endurance of more than 30 minutes, and a payload capacity of up to 6.5 pounds (2.9 kilograms).
Initial work under the partnership will focus on system integration, demonstrations, and pilot programs for US and allied defense customers.
The companies said the effort could later expand into larger-scale production and operational deployment.
Overwatch Software
SPARC AI’s Overwatch is designed to support drones and autonomous systems in maintaining positioning and targeting accuracy.
Rather than relying on satellite navigation, the software uses onboard sensors, known landmarks, and geometric calculations to determine the platform’s position and target coordinates.
This allows the system to estimate location in standard coordinates, including latitude and longitude.
A core issue for small drones is inertial measurement unit drift, in which navigation errors gradually accumulate after takeoff as onboard sensors lose accuracy over time.
Overwatch is designed to correct that drift by updating the drone’s position each time it passes a known waypoint or landmark, helping restore navigation precision without requiring GPS access.
The system is software-based and works with existing onboard sensors, cameras, and compatible visual navigation systems.
It integrates machine learning models to reduce sensor error and improve tracking and positioning accuracy while keeping onboard computing demands relatively low.
