Ann: BrainChip Evaluates Redomiciling to US, page-363

There are numerous applications for Brainchip's technologies in both edge and non-edge applications.

However, of immediate interest to the US (and other countries and defense contractors) is developing autonomous weapons (drones) that can reliably seek, identify and attack targets without communication back to command centres.

Making this identify-target-and-attack decision needs to be autonomous because in contested battle zones both sides engage in ECM (electronic counter measures) that block communications. Once the drone enters the battle zone its on its own.

This requires a number of guidance and control systems (that problem has been solved) but the target recognition system needs to be able to...
Process information from a range of sensors: optical sensors (e.g. video),LIDAR, infra-red heat signatures, active radar, radio frequency emissions etc. Terrain feature following navigation (because GPS is denied).
AI processing power: these sensors are streaming data which needs to be scanned for target profiles. For high reliability, the AI must match signatures from more than one sensor. Processing requires matching mixed signal profiles that can recognize a target (with very high reliability) in micro-seconds. That needs reasonably grunty AI and the capacity to store a wide range of profiles (training).
Light weight and small footprint: space in drones is obviously limited.
Low power draw: drones are either battery powered or powered off a generator driven by fossil fuel engines (reciprocating internal combustion, rotary (Wankel) engines, or turbines (obviously miniaturized). Either way, power draw increases battery or fuel load weight which reduces payload capacity (you want to pack the biggest explosive charge you can without reducing range or loitering time).
Low cost: Drones are mostly expendable (kamikaze drones) their principal advantage is affordable mass.

Swarming: Advanced drone technology is enabling drones to fly in large groups to overwhelm target self-defense systems. This requires swarming behavior. They are studying how birds fly in swarms and obviously how the birds communicate with each other to maintain separation and group decisions about mutual agreement on destinations. Somewhere in there is another AI workload.

All this sounds like a job description for Akida.

We are talking about potentially MILLIONS of sales units.

Excerpt from article:

In early 2024, Ukraine aimed to produce one million drones within the year. In March, that goal was revised to two million, and in October, Ukrainian president Volodymyr Zelensky announced the country was now capable of manufacturing up to four million drones annually. Russia is reportedly matching this pace of production. It is estimated that about 100 different types of drones are in use in Ukraine, ranging from toy-sized systems to larger models with wingspans of almost 20 meters.



https://ecfr.eu/article/drones-in-ukraine-four-lessons-for-the-west/


And that's just drones - what about missiles?
And that's just things that fly - what about drone boats and subs?
What about drones on wheels or tracks?
What about robotic soldiers?
And then there is cognitive radio.
And what about secret squirrel stuff - applications that have been dreamed-up but are top secret?

Wherever you look in military applications AI has a use-case.





***