Swedish startup presented an autonomous drone based on Nvidia Jetson Orin Nano

A Swedish company has developed an autonomous strike drone capable of independently identifying and hitting targets without the involvement of an operator or external communication channels. The system is based on the Nvidia Jetson Orin Nano computational module, which provides the operation of onboard machine vision algorithms.

The device uses a compact neural network to analyze visual data in real-time. The choice of the Jetson Orin Nano hardware platform is due to its low power consumption and high performance, which is critical for autonomous unmanned systems. The model is able to recognize objects in the absence of a GPS signal and electronic suppression.

A key feature of the development is the complete rejection of data transmission to external servers. All calculations are performed locally on the drone, which eliminates the possibility of intercepting control or blocking communication. The software is optimized to work with limited memory and computational resources, characteristic of embedded systems.

The technological solution demonstrates a trend towards the miniaturization of AI tools for the military sphere. The use of specialized chips makes it possible to transfer complex computer vision tasks from cloud servers directly to endpoint devices, ensuring the autonomy of mission execution in complex conditions.