We have seen that intelligent agents might take into account certain information when choosing a rational action, by including information from its sensors, information from the world, information from earlier states of the world, information from its information and goal from its utility function(s). We also need to take into account some particulars about the environment it works in. Superficially, this consideration would appear to apply more to robotic agents moving around the real world. But the considerations also apply to software agents who are receiving data and making decisions which affect the data they receive - in this case we can think of the environment as the flow of information in the data stream. i.e., an Artificial Intelligent agent may be employed to dynamically update web pages based on the requests from internet users.
In a few cases, sure aspects of an environment which should be taken into account in decisions about actions cannot be available to the agent. For instance, this could happen, because the agent cannot sense definite things. In these types of cases, we say the environment is partially inaccessible. In this type of case, the agent can have to make (informed) guesses about the inaccessible data in order to act rationally.
The builders of RHINO talk about "invisible" objects from with RHINO had to deal. These included bars and glass cases at several heights which could be undetected by the robotic sensors. These are visibly inaccessible aspects of the environment, and RHINO's designers took this into notice when designing its programs.
If we can determine what the accurate state of the world will be after an action of agent, we say the environment is deterministic. In this type of cases, the state of the world after an action is dependent only on the beginning of the world before the action and the selection of action. If the environment is non-deterministic, then utility functions will have to make guesses about the expected state of the world after possible actions if the agent is to correctly choose the best one.
RHINO's world was non-deterministic because people moved around, and they move objects likewise chairs around. Actually, visitors often tried to trick the robot by setting up roadblocks with chairs. It was another reason why RHINO's plan was frequently updated.
If an agent's present choice of action does not depend on its past actions, then the environment is called episodic. The agent will have to plan ahead, because its current action will affect subsequent ones, in non-episodic environments
Considering only the goal of getting to and from exhibits, the particular trips between exhibits can be seen as episodes in RHINO's actions. Once it had reached at one exhibit, how it got there would not in general affect its selection in getting to the next exhibit. If we also consider the goal of giving a guided tour, though, RHINO might at least remember the exhibits it had previously visited, in order not to repeat again and again itself. So, its actions were not episodic at the top level.
Static or Dynamic
An environment is static if it doesn't vary while an agent's program is making the decision about how to act. When designing agents to operate in non- static (dynamic) environments, the underlying program may have to refer to the changing environment while it anticipate or deliberates to the change in the environment in the course of the time when it receives an input and when it has to take an action.
RHINO was very speedy in making decisions. but, because of the amount of visitor movement, by the time RHINO had planned a route, that plan was sometimes wrong because someone was now blocking the route. Although, because of the speed of decision making, in its place of referring to the environment during the planning process, as we have said before, the designers of RHINO chose to enable it to continually update its plan as it moved.
Discrete or Continuous
The behaviour of the data coming in from the environment will put affect how the agent should be designed. In specific, the data can be discrete (composed of a restricted number of clearly defined parts) or continuous (seemingly without discernible sections). Of course, given the behaviour of computer memory (in bits and bytes), even streaming video may be shoe-horned into the discrete category, but an intelligent agent will possibly have to deal with this as if it is continuous. The mathematics in your agent's programs will differ depending on whether the data is taken to be or continuous and discrete.