Environments#
Grid-world environments for cooperative tasks.
Multi-agent grid environment with dynamic reward activation and penalties. |
|
Regime 0: Single-step reward at center regardless of target labels. |
|
Regime 1: Single target 'rl' activated after center. |
|
Regime 2: Single target 'ud' activated after center. |
|
Regime 3: Multiple directional targets after center. |