Consider an environment, where an agent intends to move from cell "A" to cell "G", avoiding obstacles (cells marked with shading). The agent can move forward, rotate 90º to the left, or rotate 90º to the right, and can identify the type of cell in front of him. Here's a diagram.
Classify the characteristics of the environment, according to:
- Accessible or non-accessible
- Deterministic or non‐deterministic
- Episodic or non‐episodic
- Discrete or continuous
- Static or dynamic
The environment is :
Deterministic : the future state of the environment (and the agent) is defined by the agent's action.
Partially Observable (partially accessible) : since his sensors don't see what's beyond the shaded barrier, the environment is partially observable (accessible).
Sequential (non-episodic): Since the agent's current action will constrain and affect the space of possibilities (options, choices) of future actions, current decisions are not independent of future ones.
Discrete : Environment is not continuous, that is, every action/step/case is finite (either turn left, right, forward or backward), you don't have other actions, which makes it a discrete task environment.
Static : the environment is not changing, it's static, only the agent is moving.