Agents Navigation

High-Performance Navigation Architecture

The Agents Navigation package is designed as a modular and scalable solution for managing agent movement within Unity. Built specifically with the Data-Oriented Technology Stack (DOTS) in mind, the framework utilizes Unity’s modern tech stack to maximize efficiency. This includes the use of SIMD mathematics, the C# Job System, and the Burst compiler to handle complex navigation logic. By leveraging the Entity Component System (ECS), the system is capable of handling large numbers of agents through parallelized code, ensuring high performance out of the box.

Implementation Workflows: ECS and Hybrid Modes

While the architecture is rooted in ECS, the package supports multiple development workflows to accommodate different project requirements. For developers working within a pure Entity-driven environment, the package utilizes ECS natively to maintain maximum performance.

For projects using standard Object-Oriented Programming (OOP), a hybrid workflow is available. This mode allows GameObjects to be synchronized with Entities, enabling the navigation features to be used within a traditional workflow. While this provides a lower barrier to entry for existing projects, it is noted that performance in hybrid mode may be slightly lower than a pure ECS implementation due to the overhead of synchronization.

Core Navigation and Steering Features

The framework includes a variety of navigation and spatial awareness features that work across both 2D and 3D environments. These behaviors are designed to be modular, allowing developers to add or remove them based on the specific needs of their AI. The package provides the following core features:

• Unity NavMesh Support: Includes multithreaded support for 3D NavMesh navigation.
• Avoidance: Local steering logic to prevent collisions between agents in 2D and 3D spaces.
• Flocking: Specialized 3D behaviors, specifically Alignment and Cohesion, to simulate group dynamics.
• Separation: Maintains individual agent spacing in 2D and 3D environments.
• Collision Detection: Integrated systems for handling agent-to-agent or agent-to-environment collisions.

API Structure and Extensibility

The package employs a multi-layered API that allows for low-level interactions, catering to developers who need to customize the underlying navigation logic. The codebase is built with a clear architecture that emphasizes clean coding practices, making it easier to scale and extend. Because the system is modular, it can be expanded with minimal changes to the core package.

Customization is a central focus, as the developer acknowledges that every game has unique navigation requirements. While the system is designed to be easily modified, efficient extension of its functionality typically requires a working knowledge of ECS. This modularity also extends to third-party integrations; the package currently supports out-of-the-box integration with the A* Pathfinding Project and is compatible with other extensions like Crowds, the CIVIL-AI-SYSTEM, and Animatron.

Application and Genre Versatility

The framework is intended to serve as a foundation for any project requiring robust navigation. The current version is specifically tailored for genres that demand high unit counts and precise movement, such as 2D and 3D Real-Time Strategy (RTS) games, MOBAs, RPGs, and shooters. By providing a scalable codebase, the package allows developers to build complex AI behaviors on top of a performance-optimized baseline.

Visual Breakdown