EDGE AI GPS-Denied Navigation Module for UAVs | CM4 & CM5 Integration
Introducing our EDGE AI Navigation Module — designed to enable advanced navigation capabilities for unmanned aerial systems, including continued navigation in GPS/GNSS-denied environments.
Built around Raspberry Pi Compute Module 4 (CM4) and Compute Module 5 (CM5), the module integrates with existing autopilot architectures through MAVLink while providing an onboard computing layer for navigation and autonomy applications.
The platform can support capabilities such as:
• Visual navigation
• Visual-inertial navigation (VIO)
• GPS/GNSS-denied navigation
• Computer vision and perception
• AI-based navigation algorithms
• Custom autonomy applications
A key part of the architecture is its software-defined approach. The hardware provides the computing, camera, sensor and autopilot interfaces, while additional capabilities can be developed and deployed through the software layer based on mission and platform requirements.
This allows the same EDGE AI hardware platform to evolve beyond a single navigation function without requiring the aircraft's existing flight controller to be replaced.
In this demonstration, we showcase integration with:
• Raspberry Pi Compute Module 4 & 5
• Pixhawk Cube Orange+
• Holybro Kakute H7 V1.5
• MAVLink-based autopilot communication
The module receives sensor and flight-state information from the connected autopilot, performs edge processing, and can communicate the resulting navigation information back to the flight-control system.
Interfaces include:
• 3× UART
• 3× I²C
• SPI
• Camera connectivity
• MAVLink autopilot integration
One hardware platform. Multiple software-defined capabilities.
Developed by PSYC Aerospace & Defence Industries Pvt. Ltd.
#GPSDenied #VisualNavigation #EdgeAI #UAV #AutonomousSystems #ComputerVision #MAVLink #Pixhawk #RaspberryPi #DroneTechnology