- Role
- Developer, Game designer
- Team
- Santiago Rodríguez, Marco Retana
- Technologies
- Unity, C#, Python, Mesa
An agent-based 3D simulation blending Python server logic with real-time Unity visualization.
Robbo-Harvest Rumble is a 3D simulation where a squad of robots must detect, collect, and transport mutant fruits across a grid. Behind the playful aesthetic, the project is a complex agent-based model where a Python server dictates the logic and Unity renders the environment in real time.
Server architecture
The simulation logic was built using Python and the Mesa library for agent-based modeling. The server tracks the state, position, and interactions of multiple autonomous agents simultaneously, acting as the computational brain of the simulation.
Pathfinding and constraints
A core constraint of the simulation was collision avoidance: no two agents could occupy the same coordinate. To solve this, I implemented a zone-based pathfinding system. Robots sweep designated columns in a search pattern until they detect a target. Once found, they calculate the shortest path to the deposit bin using Euclidean distance, then return to their assigned sector to resume searching.
Real-time Unity integration
The primary technical challenge was bridging the Python simulation with Unity for real-time 3D visualization. The local server streams agent coordinates to the Unity client via C#. To resolve latency and synchronization issues during asset generation, I developed a dynamic spawning system using multiple prefabs that react instantly to the server state, ensuring the 3D environment remains a perfect reflection of the underlying data model.
