Beyond Pathfinding: Making RTS Unit

Beyond Pathfinding: Making RTS Unit Movement Feel Natural in Unreal Engine 5

A real-world look at solving AI movement, group coordination, and gameplay challenges with Unreal Engine 5 and C++.

A game can compile perfectly and still feel wrong the moment you start playing it.

That's something I experienced while working on a real-time strategy (RTS) gameplay feature in Unreal Engine 5.

The challenge sounded straightforward: make groups of military units move through narrow passages without getting stuck.

But once dozens of units started moving together, things became much more interesting.

The Challenge: When Pathfinding Isn't Enough

Imagine selecting a large group of military units and ordering them through a narrow road between two structures.

Each unit knows where it needs to go. But when everyone approaches the same opening, problems start appearing.

Some units crowd around the entrance. Others wait unnecessarily, even when there's space available. A few may struggle to recover when their movement gets interrupted.

The problem isn't always the pathfinding algorithm itself.

It's how multiple units coordinate their movement while sharing limited space.

My goal was to make the movement feel natural, efficient, and believable, without sacrificing collision safety.

My Approach: Smarter Movement, Not Just Faster Units

I worked with the existing Unreal Engine C++ architecture, examining movement commands, navigation, unit states, formation assignments, and collision behaviour.

Rather than simply increasing unit speeds, I focused on improving how units approached and passed through restricted spaces.

The important areas included:

Formation movement: Keeping groups organised as they approach obstacles.

Chokepoint coordination: Managing lanes and waiting positions.

AI navigation: Using available movement space without unnecessary congestion.

Collision avoidance: Maintaining safe separation between units.

Movement recovery: Preventing units from remaining stuck after other units have passed.

One of the most important lessons was that avoiding a complete traffic jam isn't enough.

Movement should also look and feel right to the player.

Development with Unreal Engine 5, C++ and Codex

My workflow combined traditional game engineering with AI-assisted development.

I used Unreal Engine 5 and C++ for gameplay implementation, VS Code with OpenAI Codex for code analysis and iteration, and GitHub for version control and reviews.

Codex helped me investigate unfamiliar systems and make focused changes.

But the most valuable work happened during testing.

I repeatedly ran the game, watched how units behaved, identified specific problems, and refined the implementation.

A successful build didn't automatically mean the gameplay was ready.

From Development to QA Approval

Real gameplay testing revealed issues that weren't obvious from reading the code.

The movement needed to be clearer, more responsive, and better coordinated under heavy traffic.

I worked through those issues using focused debugging, repeatable gameplay scenarios, and feedback from QA.

The feature ultimately passed QA after the necessary revisions and supporting evidence were delivered.

That experience reinforced an important lesson:

Good game development isn't only about making mechanics function. It's about making them reliable, responsive, and enjoyable to use.

Building Your Next Game?

I enjoy working on the challenging parts of game development — the systems players interact with every second.

Whether you're building an RTS, action game, simulation, or another Unreal Engine project, I can help with custom C++ gameplay mechanics, AI behaviour, navigation, debugging, and performance improvements.

Have a game idea or an existing project that needs better gameplay?

Send me a message on Contra. I'd love to hear what you're building and explore how we can make it better.

Let's build something players will love.