12
Adversarial Search
12.1 Two-Agent Games
One of the challenges mentioned in Chapter 10 involved planning and acting in environments populated by other active agents. Lacking knowledge of how other agents might act, we can do little more than use a sense/plan/act architecture that does not plan too far into the unpredictable future. However, when knowledge permits, an agent can construct plans that explicitly consider the effects of the actions of other agents. Let’s consider the special case of two agents. An idealized setting in which they can take each other’s actions into account is one in which the actions of the agents are interleaved. First one agent acts, then the other, and so on.
For example, consider the grid-space world of Figure 12.1 ...
Get Artificial Intelligence now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
