Simple — Navigation
Environment
- PettingZoo:
simple_v3| Agents: 1 | Landmarks: 1
Objective
Single agent navigates to a single landmark. Tests basic spatial reasoning.
Observation Space
| Field | Dim | Description |
|---|---|---|
vel | 2 | Agent velocity [vx, vy] |
landmark_rel | 2 | Relative to landmark [dx, dy] = landmark - agent |
Reward Function
reward = -‖agent_pos - landmark_pos‖² = -(dx² + dy²)Action Space
5D: [no_op, left, right, down, up] ∈ [0,1]⁵
Run
bash
python simple.py