Read nearby trails
Each agent samples the field around it using a configurable sensor distance and angle.
Physarum
Grow living networks from thousands of simple agents
Adjust sensing, movement and trail behaviour, then watch organic structures emerge without central control.
Agents · sensing · trails · collective growth
Sense → turn → move → deposit
Physarum is an agent-based generative sandbox inspired by slime mould. Thousands of simple agents sense trails, choose a direction, move and leave a new signal. Their repeated local decisions produce complex organic networks without a central controller.
Choose a direction
The local signal guides the agent left, right or forward according to the behaviour rule.
Advance through the field
A movement distance controls how far the agent travels on each simulation step.
Leave a new trail
The agent adds signal to the field, changing what nearby agents will sense next.
What looks like life is built from logic. This simulation is not programmed to draw — it grows.
What you can explore
The system is slime-mould-inspired rather than a claim of biological realism. Its value is in seeing how sensing, movement and feedback create distinct forms of collective behaviour.
Dense networks
Use a strong, persistent trail and tightly spaced agents to fill the field with connected cellular paths.
Fine branching lines
Change sensor and movement distances to let thin paths split, reconnect and compete for space.
Waves and vortices
Adjust turning behaviour and watch local alignment develop into rotating or pulsing structures.
Symmetry and disruption
Begin with a deliberate field or paint into the running system, then observe how symmetry survives or breaks.
Unstable growth
Push sensing and trail settings toward chaotic regimes where paths continuously appear, dissolve and reform.
Collective behaviour, visible
The captures pair a real simulation output with the two compact rules behind every agent: move through the field, then sense where to turn next.
Controls and workflow
Start with a field, tune the agent loop and keep the combinations that produce interesting growth.
Choose a field
Load a saved system or begin with a new agent distribution.
Tune perception
Adjust sensor distance and the angles agents use to read trails.
Shape behaviour
Change turn angle, movement distance and trail strength.
Paint into the field
Intervene by touch while the simulation grows around the new signal.
Save and restore
Store interesting agents, fields and rules for later exploration.
Emergent Systems Sandboxes
These projects explore the same central idea at different scales: complex behaviour can emerge from many simple local rules.
I build tools for exploring how complex behaviour emerges from simple rules.
See all projects