For each run in cooperative navigation, the environment begins from new (random) starting locations for the agents and landmarks. This causes the learning curves to start from different positions in the graph. Need to make sure that each time we train, we must begin from the same starting point.
For each run in cooperative navigation, the environment begins from new (random) starting locations for the agents and landmarks. This causes the learning curves to start from different positions in the graph. Need to make sure that each time we train, we must begin from the same starting point.