// 3x3 grid
// based on Littman, Cassandra and Kaelbling
// Learning policies for partially observable environments: Scaling up  
// Technical Report CS, Brown University

pomdp



// only the target is observable which is in the south east corner
observables
	o
endobservables

module grid
	
	x : [0..2]; // x coordinate
	y : [0..2]; // y coordinate
	o : [0..2]; // observables
	// 0 - initial observation
	// 1 - in the grid (not target)
	// 2 - observe target
	
	// initially randomly placed within the grid (not at the target)
	[] o=0 -> 1/8 : (o'=1) & (x'=0) & (y'=0)
			+ 1/8 : (o'=1) & (x'=0) & (y'=1)
			+ 1/8 : (o'=1) & (x'=0) & (y'=2)
			+ 1/8 : (o'=1) & (x'=1) & (y'=0)
			+ 1/8 : (o'=1) & (x'=1) & (y'=1)
			+ 1/8 : (o'=1) & (x'=1) & (y'=2)
			// + 1/8 : (o'=1) & (x'=2) & (y'=0) the target
			+ 1/8 : (o'=1) & (x'=2) & (y'=1)
			+ 1/8 : (o'=1) & (x'=2) & (y'=2);
			
	// move around the grid
	[east] o=1 & !(x=1 & y=0) -> (x'=min(x+1,2)); // not reached target
	[east] o=1 & x=1 & y=0 -> (x'=min(x+1,2)) & (o'=2);
	[west] o=1 -> (x'=max(x-1,0)); // not reached target
	[north] o=1 -> (y'=min(y+1,2)); // reached target
	[south] o=1 & !(x=2 & y=1) -> (y'=max(y-1,0)); // not reached target
	[south] o=1 & x=2 & y=1 -> (y'=max(y-1,0)) & (o'=2); // reached target
	
	// reached target
	[done] o=2 -> true;
	
endmodule

// reward structure for number of steps to reach the target
rewards
        [east] true : 1;
        [west] true : 1;
        [north] true : 1;
        [south] true : 1;
endrewards

// target observation
label "goal" = o=2;