2.6 Implement a table-lookup agent for the special case of the vacuum-cleaner world consisting of a 2 x 2 grid of open squares, in which at most two squares will contain dirt. The agent starts in the upper left corner, facing to the right. Recall that a table-lookup agent consists of a table of actions indexed by a percept sequence. In this environment, the agent can always complete its task in nine or fewer actions (four moves, three turns, and two suck-ups), so the table only needs entries for percept sequences up to length nine. At each turn, there are eight possible percept vectors, so the table will be of size 89 = 134,217,728. Fortunately, we can cut this down by realizing that the touch sensor and home sensor inputs are not needed; we can arrange so that the agent never bumps into a wall and knows when it has returned home. Then there are only two relevant percept vectors, ?0? and ?!?, and the size of the table is at most 29 = 512. Run the environment simulator on the table-lookup agent in all possible worlds (how many are there?).

Record its performance score for each world and its overall average score.