Artificial Life that learns Foraging Behavior using Neuroevolution

Agents exhibit foraging behavior via neural networks that are generated using an evolutionary algorithm.

The green blocks are targets in space representing ‘food’, which the agents seek and collect.

The agent with the highest score (that is, the most amount of food collected over time) is outlined by a red box.

Each agent recieves inputs regarding the relative distance and angle between itself and the nearest target at any given point within a certain area around itself.

If an agent has no target within range, it takes a random nearby position as its target allowing for fidgeting behavior that can potentially move it toward a valid target.

The red lines extending from each agent show which target is currently providing inputs.

The red text over each agent displays their numerical ID along with their current spot on the leaderboard.

The position and angle of each agent is randomized at the start of each new trial.

When all targets have been collected in a given trial, the space is reset with a random distribution of new targets.