a)
The robot has no knowledge about the rest of the grid and its neigbouring squares.
It always acts rationally in its present square i.e sucks if there is dirt and moves if the square is clean.

b)
The greedy robot is non-deterministic when it has more than one choice to move away from its present square.
This is because in case of a tie,it randomly chooses a neighbour(from among those which are involved in tie).

c)
It is not possible to design an environment in which we can say for sure that the randomized agent
will always perform poorly.One of these random paths is the optimal path.

Consider this environment:
0.7 0.4 0.5 0.8 0.3
0.5 0.5 0.7 0.2 0.6
0.9 0.1 [0] 0.4 0.1
0.0 0.0 0.7 0.0 0.0
0.0 0.0 0.1 0.0 0.0
0.0 0.0 0.0 0.0 0.0

Suppose the no of moves the robot is allowed to make is small (say 5)
There are two choices for the robot in its present situation.If it moves up,it will perform well but
if it moves down,it will perform poorly. 

d)
If the sensing is noisy,then the lower_bound(a variable in the code of part C) will also deviate by
20%.The new lower_bound will be either  1.2*lower_bound or 0.8*lower_bound(choose randomly).

e)
Since the robot does not keep track of its traversed path,its behaviour will not 
be affected if the cleaned squares become dirty.