DECISION TRACE
LIVE 3D AUTONOMOUS SYSTEMS TEST
Evolutionary robot laboratory.
A seeded optimizer creates one compact challenger. Six autonomous agents then plan around obstacles with A*, select targets, evade, and fire simulated energy pulses. Nothing here controls physical hardware.
What is actually computed
Robot parameters, fitness trials, navigation paths, line of sight, movement, hits, damage, and the winner are computed in your browser. Reusing a seed reproduces the initial design and arena.
MEASUREMENT BOUNDARY
Inspect, don’t imagine.
- Designer: seeded evolutionary search, not an LLM claim.
- Navigator: grid-based A* with obstacle occupancy and replanning.
- Controller: utility rules choose pursue, evade, or fire from current state.
- Combat: fictional energy pulses with line-of-sight collision and health accounting.
- Renderer: vendored Three.js 0.180.0; simulation state remains independent of frame rate.