Loyal Wingmen with Attitude: The Rise of Smarter, Deadlier Drones in Modern Warfare

7 min readNov 11, 2024

So, here we are, folks. We’re talking about drones with brains now — yeah, because just having mindless flying robots wasn’t enough. No, no, we needed to pump these things full of Deep Reinforcement Learning (or DRL if you’re hip to the tech lingo). And now, these drones don’t just fly around like some mindless robotic birds. No, now they learn, they adapt, they make decisions. They’re practically having board meetings in the sky about which target to take out next. And they call this little circus “Loyal Wingman.” Loyal! Because if there’s one thing I want in my autonomous weapon, it’s loyalty. Makes you feel all warm and fuzzy, doesn’t it?

So, What’s This Deep Reinforcement Learning All About?

Think about it: we’re taking artificial intelligence and deep neural networks and throwing them into the skies. These UAVs (that’s Unmanned Aerial Vehicles for the rest of us) don’t just follow some preprogrammed routine. No, these guys are out there, learning. They’re figuring out how to fly, fight, dodge, and occasionally duck out of a firefight when things get a little too hot. They run a million scenarios, fail, succeed, try again, and eventually, they’re pulling maneuvers that would make Maverick look like a beginner. It’s like turning air combat into a giant video game for AI — and these drones keep leveling up.

And here’s the kicker: once they’ve learned a trick, they don’t forget it. These aren’t your Uncle Marty’s drones from the ’90s. These things get smarter every time they’re up in the sky, figuring out how to outmaneuver whatever you throw at them. They’re like the “yes men” of warfare, right up until they take over the meeting and tell the humans what to do.

The Brains of the Operation: Key Components of DRL-Based Air Combat

To pull this off, these drones need a pretty sophisticated set of tools. It’s not as simple as “fly here, bomb there.” No, no, it’s more like:

State Representation:

You ever think about how you know where you are without looking at a map? These drones are doing that, constantly, but they don’t have eyes. They’re using data — position, velocity, where the enemy is, where the trees are, where they left the keys — all of it. And they need to keep this model up-to-date in real time, or else someone’s gonna end up with a multi-million-dollar drone up a tree.

Action Space:

Drones have all these moves they can pull — turns, rolls, dives, missile launches. This is the drone’s version of a Swiss Army knife, and DRL figures out which action to pull out and when. This means engineers have to map out every possible move like some celestial dance routine, so the drone can make the right call at the right moment.

Reward Function:

This is where it gets funny: drones don’t just fight to survive, they’re chasing rewards! Yep, they’re like dogs after treats. Only their “treat” is winning the fight, or dodging a missile, or just not blowing up today. The better they perform, the more rewards they get, and the more their “behavior” improves. That’s right, drones have behavior now — next thing you know, they’ll have personalities.

Policy and Value Networks:

These are like the drone’s playbook. One network says, “Here’s what you do next,” and the other one says, “And here’s why.” They’re constantly weighing their options to pick the best move based on what’s happening right now and what’s likely to happen a few seconds down the line. So yeah, we’re talking about robots with instincts. Welcome to the future, folks.

But, Hey, It’s Not All Smooth Flying

If only it were that easy! Getting these flying death machines to work isn’t just a walk in the park.

High-Dimensional State and Action Spaces:

These drones are handling so much data that even a computer would sweat. They’re tracking the enemy, watching the sky, checking their ammo, and figuring out where to turn next, all at once. To make it work, they need the tech equivalent of an Olympian brain.

Sparse and Delayed Rewards:

Here’s a kicker: sometimes these drones don’t know they did a good job until ten minutes later. Imagine playing a video game where you don’t find out if you scored until halfway through the next level. It’s hard to learn anything that way! But engineers have some fancy tricks, like predictive modeling, that help these things figure out the value of their moves, even if the payoff is delayed.

Safety and Reliability:

Let’s be honest — these are multi-million-dollar autonomous fighter jets. Losing one isn’t like dropping your phone. You want these drones to follow the rules, not pull a wild stunt and go down in flames. Researchers are working on ways to keep these things “safe” and predictable, or as close as they can get.

New Tricks and Tactical Magic

So, what can these DRL drones actually do? Turns out, plenty.

Autonomous Tactical Maneuvering:

Forget simple left turns. These drones can pull off complex maneuvers on their own: high-speed turns, evasive rolls, you name it. Basically, they’ve got an entire catalog of moves to play with, and they don’t need a human yelling at them to pull it off.

Swarm Coordination:

Here’s where it gets scary. One drone? That’s fun. Ten drones? That’s a swarm, people. And DRL lets these drones team up, like a tactical hive mind. One plays decoy, another attacks, and the rest gather intel or flank the enemy. It’s like a high-flying flash mob of artificial intelligence.

Predictive Modeling:

As if they didn’t already have enough, these drones are also learning to predict enemy behavior. They watch, they learn, and they can anticipate moves before they happen. So now we have drones that read minds… or at least close enough to it.

Enhanced Human-Machine Teaming:

Here’s the grand finale: these drones don’t just fly around on their own. They work with human pilots. They’re the wingman that doesn’t ask questions, doesn’t make mistakes (well, we hope), and can take orders without any attitude. Pilots can focus on the mission while these drones handle the grunt work, like dodging bullets or gathering data.

Looking Down the Barrel of the Future

But, hey, with great power comes great… complications. As these drones get smarter and more autonomous, a few issues start to crop up.

Ethics of Autonomy:

You want to talk ethics? Try explaining to a drone that it’s not okay to take out a civilian because they’re standing in the wrong spot. Teaching a machine the difference between right and wrong is like teaching calculus to a hamster. We’re still figuring out how to make sure these things don’t get trigger-happy.

Global Power Shifts:

Here’s a cheery thought: as more countries get their hands on these little Loyal Wingmen, the power balance could shift. Entire fleets of AI-driven drones buzzing around with little human oversight. It’s a recipe for, well, a lot of fun at the United Nations.

Human-in-the-Loop Training:

Some engineers are thinking, “Hey, maybe let’s keep a human around to keep an eye on things.” Because if there’s one thing we’ve learned, it’s that even the smartest robots need someone to remind them to follow the rules.

Final Thought

So here we are, folks, staring down the barrel of the future with drones that not only follow commands but learn, strategize, and adapt. Deep Reinforcement Learning has turned these flying machines into something that sounds like it belongs in a sci-fi movie. The “Loyal Wingman” is more than just a fancy buzzword — it’s a reality. These drones aren’t just along for the ride. They’re running the show, one maneuver at a time.

But as we send these artificial pilots into the skies, we’d better keep our eyes open. Because with technology this powerful, there’s a fine line between progress and just plain asking for trouble. So here’s to the future, folks — hope you packed a parachute.

If you’re enjoying the content on my blog and would like to dive deeper into exclusive insights, I invite you to check out my Patreon page. It’s a space where you can support my work and get access to behind-the-scenes articles, in-depth analyses, and more. Your support helps me keep creating high-quality content and allows me to explore even more exciting topics. Visit [patreon.com/ChristianBaghai](https://www.patreon.com/ChristianBaghai) and join the community today! Thank you for being a part of this journey!

Christian Baghai | Patreon

Switzerland: The Secret Safe Haven for Organized Crime Beneath a Surface of Stability | Patreon

The Defence and Intervention Frigate (FDI): It’s a Warship, Not an IT Department | Patreon

The Defence and Intervention Frigate (FDI): A Warship That Thinks It’s a Server Farm | Patreon

The Defence and Intervention Frigate (FDI): When a Warship Thinks It’s an App Store | Patreon

BFMTV’s Twitter Strategy: News or Just Outrage? | Patreon

Trump’s Trap: The Perils of Strongman Politics and the Path to a Forever War | Patreon

Understanding the Divide: How Trump Supporters React to Losing Versus, You Know, “Normal” People | Patreon

DJT Stock: The Roller Coaster That Forgot It’s a Stock | Patreon

How to Break Russia’s Propaganda Spell: A Plan So Clear Even They Might Censor It | Patreon

Flying High in a War Zone: Ukraine’s Bold Plan to Reopen Its Skies Amid Conflict | Patreon

Royals, Rage, and Real-Time Intel: How OSINT Could Have Saved Spain’s King from a Muddy Reception | Patreon