Primeordial Labs controls military drones with natural language

Primeordial Labs wants to take the stress out of drone operators on the battlefield. They are to control drones simply using natural language.

listen Print view
Operator in the foreground controls drone behind him by voice.

An operator controls the drone in the background using only natural language.

(Image: Primeordial Labs (Screenshot))

3 min. read

With Anura, the US security company Primeordial Labs has developed software that is set to revolutionize the control of drones on the battlefield. Instead of having drone missions corrected manually by a controller using a control stick, the drones are supposed to obey the word and also carry out complicated maneuvers in response to voice instructions. Initial tests prove that this works, writes Defense News.

The guidance of drones during missions is partly automatic and partly controlled manually by operators. However, the software of the ground control station does not always work as desired. Lee Ritholtz, CEO and co-founder of Primeordial Labs, knows what he is talking about. He worked on such software himself for many years, including at the defense companies Lockheed Martin and Sikorsky. Today, he describes the ground control station control software of the time as “corrosive”.

Working with machines in a mission is very stressful and quickly leads to overload because the operator has to physically realize his ideas in his head first. This takes time, which is not available. Ritzholtz's aim was therefore to make interaction with military drones more human and to minimize stress and overload.

The result is the Anura control software, which can respond to natural language input and execute commands. It is true that there was already software that could be used to transmit voice commands to drones. However, keywords and phrases were used for this purpose. However, this is not conducive to having drones carry out complex maneuvers quickly. In the worst-case scenario, military missions could even fail as a result.

The operator's sentences are broken down into a sequence of instructions via Anura. Anura is supposed to understand the underlying intention. A certain form of autonomy is also built in. The software receives feedback on the status of the implementation of the commands and makes changes if necessary. During the test, entire swarms of drones were also controlled by voice to complete tasks together. For example, a single operator can control several drones in military operations.

Primeordial Labs promises that the software will be able to work with any platform and any system.

During a demonstration of the system in a park in Leesburg in the US state of Virginia, Anura was used to control Skydio drones and those from Teal Drones by voice. Both drone systems are already being used for reconnaissance purposes in the U.S. Army's Short Range Reconnaissance Program.

In the demonstration, the operator communicated with the drone via a headset. The instructions were given using natural language, which the drones implemented. Mechanical control was no longer required. The drones were able to climb to certain heights, fly precisely to targets, and perform flight maneuvers to observe a target.

Videos by heise

According to Ritholtz, the system runs solely on local resources. No large language models are required, nor is access to open artificial intelligence (AI) via servers.

Primeordial Labs already has the U.S. Army as a customer. They have ordered 8,000 Anura licenses for the year 2025 to test the human-machine interface.

(olb)

Don't miss any news – follow us on Facebook, LinkedIn or Mastodon.

This article was originally published in German. It was translated with technical assistance and editorially reviewed before publication.