foundations of computational agents
An agent could assume there is no uncertainty, or it could take uncertainty in the domain into consideration. Uncertainty is divided into two dimensions: one for uncertainty from sensing and one for uncertainty about the effects of actions.
In some cases, an agent can observe the state of the world directly. For example, in some board games or on a factory floor, an agent may know exactly the state of the world. In many other cases, it may only have some noisy perception of the state and the best it can do is to have a probability distribution over the set of possible states based on what it perceives. For example, given a patient’s symptoms, a medical doctor may not actually know which disease a patient has and may have only a probability distribution over the diseases the patient may have.
The sensing uncertainty dimension concerns whether the agent can determine the state from the stimuli:
Fully observable means the agent knows the state of the world from the stimuli.
Partially observable means the agent does not directly observe the state of the world. This occurs when many possible states can result in the same stimuli or when stimuli are misleading.
Assuming the world is fully observable is a common simplifying assumption to keep reasoning tractable.
A model of the dynamics of the world is a model of how the world changes as a result of actions, or how it changes even if there is no action. In some cases an agent knows the effects of its action. That is, given a state and an action, the agent can accurately predict the state resulting from carrying out that action in that state. For example, a software agent interacting with the file system of a computer may be able to predict the effects of deleting a file given the state of the file system. However, in many cases, it is difficult to predict the effects of an action, and the best an agent can do is to have a probability distribution over the effects. For example, a teacher not know the effects explaining a concept, even if the state of the students is known. At the other extreme, if the teacher has no inkling of the effect of its actions, there would be no reason to choose one action over another.
The dynamics in the effect uncertainty dimension can be
deterministic when the state resulting from an action is determined by an action and the prior state, or
stochastic when there is only a probability distribution over the resulting states.
This dimension only makes sense when the world is fully observable. If the world is partially observable, a stochastic system can be modeled as a deterministic system where the effect of an action depends on some unobserved feature. It is a separate dimension because many of the frameworks developed are for the fully observable, stochastic action case.