AI/Game Development Blog

Developing future Artificial Intelligence programming techniques for games & devices

Void/Runner

Void/Runner is a project developed at RMIT, worked on with JW.

The game was a project to examine different methods of AI controllers,  using flocking behaviours, state machines, evolutionary neural networks and A* pathfinding to develop an engaging and intelligent game. It was developed in Unity3D, using heavily modified UnitySteer libraries to aid with steering behaviours, and a custom evolutionary algorithm for neural networks developed for the project, dubbed UnityEvolve.

Controls:
W,A,S,D to move the camera around. Click to move the ship.
Q launches fighters, SPACE turns the guiding beacon on/off, E launches an EMP.

 

Setting

VOID/RUNNER is set in the distant future, deep in the dark depths of space. The player controls the flagship of a fleet of defenceless transports carrying refugees to a distant starport. The fleet is under attack by Marauders, a merciless hunting pack of space reavers, headed by a ruthless overlord. It is up to the player to both lead and protect the fleet from untimely destruction or worse.

Gameplay

The goal of the game is to escort the fleet to the boundaries of the protected starport. This has to be done while protecting the fleet from attacks by the Marauders, and only the player has the ability to ward them off. The player will lead the fleet by turning on a beacon which becomes a target for the fleet to follow. Without a beacon the fleet will flock together but travel without a purposeful direction.

The player has a point defence laser system capable of destroying enemy ships within seconds, and a larger, limited use EMP capable of taking out a large portion of enemies instantly. A downside of using the EMP is that it will make the fleet flee the player for a time as it creates a false reading of danger on their radar. This will scatter the herd making it harder to protect them from any remaining enemies.

As leading the fleet with the beacon means that the player must position itself in front of the fleet, the player must balance defence and progress if they want to make it to safety. The Overlord will use devious tactics to try to separate the fleet for its Marauders to prey on.

The player also has a squadron of AI drone fighters which will autonomously hunt down and destroy enemy ships. These ships are controlled by evolved neural networks, which are trained to fight against the waves of marauders.

Flocking

Basic flocking

The transport ships are attracted to the transport ship as long as the beacon is switched on. They have steering behaviours for alignment, cohesion, separation, target and avoidance. They also steer to avoid obstacles.

Enemies

Enemies swarming beyond vision radius

The reavers are have basic individual controllers as well as an overmind controlling tactics to be used against the player. In this image the AI is performing a “Surround” maneuver, circling the transports.

Pathfinding

Pathfinding

If a direct path to the target destination is blocked by a collision object, all units have a* path planning which allows them to navigate complex level designs.

One ResponseLeave one →

  1. Void/Runner v0.01 | AI GameDev Blog

Leave a Reply