A while ago, I wrote a minesweeper AI. I intended to publish a writeup, but due to university and life and exams, I never got around to writing it. But having just finished my Fall term, I have some time to write a decent overview of what I did.

Short 30 second video of the AI in action here:

### How to Play Minesweeper

If you’re an experienced minesweeper player, you can probably skip this section. Otherwise, I’ll just give a quick overview of some basic strategies that we can use to solve an easy minesweeper game.

We start with a 10×10 Beginner’s grid, and click on a square in the middle:

We can quickly identify some of the mines. When the number 1 has exactly one empty square around it, then we know there’s a mine there.

Let’s go ahead and mark the mines:

Now the next strategy: if a 1 has a mine around it, then we know that all the other squares around the 1 *cannot* be mines.

So let’s go ahead and click on the squares that we know are not mines:

Keep doing this. In this case, it turns out that these two simple strategies are enough to solve the Beginner’s grid:

**Roadmap to an AI**

All this seems easy enough. Here’s what we’ll need to do:

**Read the board**. If we use a screenshot function, we can get a bitmap of all the pixels on the board. We just need to ‘read’ the numbers on the screen. Luckily for us, the numbers tend to have different colors: 1 is blue, 2 is green, 3 is red, and so on.**Compute**. Run the calculations, figure out where the mines are. Enough said.**Click the board**. This step is easy. In Java, we can use the Robot class in the standard library to send mouse clicks to the screen.

**Reading the Field**

There’s not a whole lot to this step, so I’m going to skim over it quickly.

At the beginning of the run, while we have a completely empty grid, we invoke a **calibration routine** – which takes a screenshot and looks for something that looks like a Minesweeper grid. Using heuristics, it determines the location of the grid, the size of a grid square, the dimensions of the board, and things like that.

Now that we know where the squares are, if we want to read a square, we crop a small section of the screenshot and pass it to a **detection routine**, which looks at a few pixels and figures out what’s in the square.

A few complications came up in the detection routine:

- The color for the number 1 is very close to the color of an unopened square: both are a dark-blue color. To separate them apart, I compared the ‘variance’ of the patch from the average color for the patch.
- The color for 3 is identical to that for 7. Here, I used a simple edge-detection heuristic.

**Straightforward Algorithm**

The trivially straightforward algorithm is actually good enough to solve the beginner and intermediate versions of the game a good percent of the time. Occasionally, if we’re lucky, it even manages to solve an advanced grid!

When humans play minesweeper, we compete for the *fastest* possible time to solve a grid of minesweeper. So it doesn’t matter if we lose 20 games for every game we win: only the wins count.

This is clearly a silly metric when we’re a robot that can click as fast as we want to. Instead, we’ll challenge ourselves with a more interesting metric:

**Win as many games as possible.**

Consider the following scenario:

Using the straightforward method, we seem to be stuck.

Up until now, whenever we mark a square as having a mine or safe, we’ve only had to look at a single 3×3 chunk at a time. This strategy fails us here: the trick is to employ a **multisquare** algorithm – look at multiple different squares at once.

From the lower 2, we know that one of the two circled squares has a mine, while the other doesn’t. We just don’t know which one has the mine:

Although this doesn’t tell us anything right now, we can combine this information with the next 2: we can deduce that the two yellowed squares are empty:

Let’s click them to be sure.

And voilà. They’re empty. The rest of the puzzle can be solved easily, after we’ve made the deduction that those two squares were empty.

**The Tank Solver Algorithm**

It’s difficult to make the computer think deductively like we just did. But there is a way to achieve the same results, without deductive thinking.

The idea for the Tank algorithm is to **enumerate** **all possible** configurations of mines for a position, and see what’s in common between these configurations.

In the example, there are two possible configurations:

You can check for yourself that no other configuration could work here. We’ve deduced that the one square with a cross *must* contain a mine, and the three squares shaded white below *must not* contain a mine:

This works even better than human deduction!

We always try to apply the simple algorithm first, and only if that gets us stuck, then we bring in the Tank algorithm.

To implement the Tank algorithm, we first make a list of **border tiles**: all the tiles we aren’t sure about but have some partial information.

Now we have a list of border tiles. If we’re considering every possible configuration, there are of them. With backtracking, this number is cut down enough for this algorithm to be practical, but we can make one important optimization.

The optimization is **segregating** the border tiles into several disjoint regions:

If you look carefully, whatever happens in the green area has no effect on what happens in the pink area – we can effectively consider them separately.

How much of a speedup do we get? In this case, the green region has 10 tiles, the pink has 7. Taken together, we need to search through combinations. With segregation, we only have : about a 100x speedup.

Practically, the optimization brought the algorithm from stopping for several seconds (sometimes minutes) to think, to giving the solution instantly.

**Probability: Making the Best Guess**

Are we done now? Can our AI dutifully solve any minesweeper grid we throw at it, with 100% accuracy?

Unsurprisingly, no:

One of the two squares has a mine. It could be in either, with equal probability. No matter how cleverly we program our AI, we can’t do better than a 50-50 guess. Sorry.

The Tank solver fails here, no surprise. Under exactly what circumstances does the Tank algorithm fail?

If it failed, it means that for **every** border tile, there exists **some** configuration that this tile has a mine, and **some** configuration that this tile is empty. Otherwise the Tank solver would have ‘solved’ this particular tile.

In other words, if it failed, we are forced to guess. But before we put in a random guess, we can do some more analysis, just to make sure that we’re making the **best guess** we could make.

Try this. What do we do here:

From the 3 in the middle, we know that three of them are mines, as marked. But marking mines doesn’t give us any *new information* about the grid: in order to gain information, we have to uncover some square. Out of the 13 possible squares to uncover, it’s not at all clear which one is the best.

The Tank solver finds 11 possible configurations. Here they are:

Each of these 11 configurations should be equally likely to be the actual position – so we can assign each square a **probability** that it contains a mine, by counting how many (of the 11) configurations does it contain a mine:

Our best guess would be to click on any of the squares marked ‘2’: in all these cases, we stand an 82% chance of being correct!

**Two Endgame Tactics**

Up until now, we haven’t utilized this guy:

The mine counter. Normally, this information isn’t of too much use for us, but in many endgame cases it saves us from guessing.

For example:

Here, we would have a 50-50 guess, where two possibilities are equally likely.

But what if the mine counter reads 1? The 2-mine configuration is eliminated, leaving just one possibility left. We can safely open the three tiles on the perimeter.

Now on to our final tactic.

So far we have assumed that we only have information on a tile if there’s a number next to it. For the most part, that’s true. If you pick a tile in some distant unexplored corner, who knows if there’s a mine there?

Exceptions can arise in the endgame:

The mine counter reads 2. Each of the two circled regions gives us a 50-50 chance – and the Tank algorithm stops here.

Of course, the middle square is safe!

To modify the algorithm to solve these cases, when there aren’t that many tiles left, do the recursion on *all* the remaining tiles, not just the border tiles.

The two tricks here have the shared property that they rely on the mine counter. Reading the mine counter, however, is a non-trivial task that I won’t attempt; instead, the program is coded in with the total number of mines in the grid, and keeps track of the mines left internally.

**Conclusion, Results, and Source Code**

At this point, I’m convinced that there isn’t much more we could do to improve the win rate. The algorithm uses every last piece of information available, and only fails when it’s provably certain that guessing is needed.

How well does it work? We’ll use the success rate for the advanced grid as a benchmark.

- The naïve algorithm could not solve it, unless we get very lucky.
- Tank Solver with probabilistic guessing solves it about 20% of the time.
- Adding the two endgame tricks bumps it up to a 50% success rate.

Here’s proof:

I’m done for now; the source code for the project is available on Github if anyone is inclined to look at it / tinker with it:

What version of Minesweeper is that? Windows 7?

Have you tried finding a version of Minesweeper online (say, in Flash or HTML5) and porting your AI to that version so that people can experiment with it outside of Windows? Alternatively, do any commenters feel like *writing* a Flash or HTML5 Minesweeper clone?

Holy cow, dude, just get a copy of Windows.

To be honest, it should be fairly trivial to write a minesweeper clone – I can’t imagine anything in it being particularly complicated. I would do it myself, but I’ve currently got enough things on the go.

Hey, what’s you talking about?

$ apt-cache search minesw

freesweep – text-based minesweeper

gnomine – popular minesweeper puzzle game for GNOME

kmines – minesweeper game

sgt-puzzles – Simon Tatham’s Portable Puzzle Collection – 1-player puzzle games

xbomb – ‘minesweeper’ game with squares, hexagons or triangles

xdemineur – Yet another minesweeper for X

Not everyone uses a Debian based Linux system, but he might be able to search for the sources to a few of those.

And not everybody deserves to get whatever program he wants for his obscure whatever OS.

If he is not using Windows, a popular Linux distro, OS X or FreeBSD, then he should go and find the game himself.

And it’s not like parent comment meant just “here are apt-get binaries of the game”. It was more in the spirit of “hey, dude, tons of bloody OSS implementations exist”.

I’m aware of freeware Minesweeper clones. However, I wouldn’t necessarily assume that the OP’s code will “just work” when presented with those UIs. And if the OP switched to using something like gnomine, then he’d be shutting out the readers who use *Windows*.

Hence, my suggestion that the OP (a) clarify which version[s] of the Minesweeper UI he was using, and (b) consider switching to a more portable UI, such as one written in Flash or HTML5.

I think a lot of the commenters here haven’t realized that the OP’s code works by screen-scraping a specific UI, as opposed to… um… magic? :P

This entire comment is erroneous and irrelevant. The OP is writing an article on the theory and process behind building a Minesweeper AI, and is *clearly* providing the source code as an example of how the algorithms and theory explained could be implemented in code.

At no point is he “shutting out” any readers, or did he claim that it would “just work” for anybody. The service he is providing is not writing a program that you can use, but teach you the process behind making an AI for minesweeper.

If you want to use a minesweeper AI, and not use the minesweeper client for which the OP’s code is *written for*, then stop asking someone to do it for you and WRITE YOUR OWN.

Here ya go: http://benru.nl/mines.html All open source, written in javascript and html.

I wrote that like 6 years ago, so the code is horrible and it’s not tested on all browsers (but it does work on the newest Opera). It does the trick though and looks exactly like the xp minesweeper. You can alter the graphics yourself by just changing the gifs.

Could you speed it up by not marking flags on the squares? I always play without using flags. The last click on the last empty box will auto-place flags on all mines and win the game.

If you middle click on a number that already has the correct amount of marked bombs around it, it will open the rest of the squares touching that number. This is just about the only way to get a reasonable time.

Pointless when you’re writing an AI though.

that’s not ai, that’s just an algorithm. you lied to me

ai IS just an algorithm …

Er, AI is all about algorithms.

And especially “Game AI” refers to exactly this kind of algorithms.

What did you expect? Thinking robots?

Are you a robot?

I have done much the same thing. However for the first move I always choose a corner square. It’s just as likely to contain a bomb as any other square, but as it only has 3 neighbors there is more chance of it not being adjacent to a bomb. So then the board opens up a little (or sometimes a lot).

Also, your 11 possible configurations leads to the wrong probabilities. The right column which reads 2,7,2 is completely wrong. Without using the total number of bombs left, those three squares each have a 1 in 3 chance of containing a bomb.

Your 11 possible configurations range from 7 to 9 flags. This by itself shows that you can’t just add up the flags and expect it to produce the correct odds.

Of the 11 combinations, two have 7 flags, three have 9 flags and six have 6 flags. If you know how many bombs are going to fit into these 16 ‘border’ squares, then by all means assign each combination an equal probability. But if you don’t know then you can’t do it that way.

Sorry, I meant that six have 8 flags. So over half the solutions have 8 flags, but your logic assumes that 7 flags is just as likely as 8. This is where I disagree – you can’t know that.

Actually you are giving more weight to solutions with 8 flags than to those with 7 or 9.

If we assign a probability of a bomb to each of the vacant squares (starting from top left and moving counter-clockwise), I would use these percentages (rounded):

67, 67, 67, 67, 40, 40, 33, 33, 20, 33, 33, 33, 33.

Which suggests there is a single best move. And none of your 6 best moves have an 82% success rate. I find one square with an 80% chance.

Sheesh, 1kSLC+ in a single .java file :(

@Bomb:

You’re right that not all 11 possibilities have the same probability of occurring — in order to give completely correct probabilities we do need to take into account the number of mines left. We do this in the second endgame tactic stage, because it’s more computationally feasible then.

I’m not sure how you got your probabilities. In the post, I never told you how many mines were left on the board, so I’m not sure how you could do a more accurate calculation without this information. (I think I set it to 200 mines to take the screenshot)

Also, on Windows 7 minesweeper, the first click always opens up its surrounding 3×3, while on XP you’re only guaranteed the one square been open.

How exactly did you determine the environment from screen shot. For example the grid size and so on. Was it simple as if screenSize.width == knownWidthForEasyVersion then create array with easy version size and so on?

No, there’s some crude computer imaging. It takes a screenshot and looks for ‘crosses’ at right angles (ie, the crosses formed by lines separating squares). Once it’s found the crosses, it works backward to deduce the size and position of the grid.

Nice at least Im not the only one to do this. I did it way back in 1995ish when I got annoyed with a crappy mouse. My version worked on win3.11 win 95 win 98, and NT. sadly it doesn’t work on vista or win 7. The biggest problem with it is now that my AI can win beginner in under 1 second and expert in under 10 seconds I never play it anymore…. :(

[...] that will take over our holiday free time. [Bai Li] just published an excellent article about writing a program that can automatically solve the game of Minesweeper. For those of you who are unfamiliar, Minesweeper gives you a grid in which land mines have been [...]

[...] that will take over our holiday free time. [Bai Li] just published an excellent article about writing a program that can automatically solve the game of Minesweeper. For those of you who are unfamiliar, Minesweeper gives you a grid in which land mines have been [...]

How fast are the best times?

[...] that will take over our holiday free time. [Bai Li] just published an excellent article about writing a program that can automatically solve the game of Minesweeper. For those of you who are unfamiliar, Minesweeper gives you a grid in which land mines have been [...]

Great algorithm, thanks for sharing this.

Reblogged this on Of_Drifter.

Heh, I did the exact same thing 12 years ago. Note that the probabilities are perfectly computable across the entire grid, so you can consider a probabilistic solver to be “perfect” even though it may not be right all the time.

I noticed your Straightforward Algorithm is basically a constraint satisfaction problem. I had to make a minesweeper AI as well and discovered that winning expert was achievable to ~33.33%

Reblogged this on Andrew Rauh.

[...] And here is the link on {Lucky’s} documentation on the algorithm. [...]

Congratulations!

I did the same about 15 years ago (I was young, and in Army, and very bored), complete with graphic-detection routine and the optimal calculations.

The result was terrific and fun – for about 1 day.

I beat everybody’s scores. More specifically, the program did. I was very proud.

I had to update the graphic detection routine when the next Windows version came out. I did not bother with the one after that.

That is not “AI”, though. There is no “self” intelligence in the program; it only makes the calculations that you have programmed into it.

It looks very intelligent to the outsiders … but that’s not what we call “real intelligence” :) :)

[...] Li’s How to Write your own Minesweeper AI describes an algorithm (with Java download) that achieves a 50% success rate at the popular Windows [...]

[...] Fast forward to just last month, just before Christmas, when I checked Reddit and saw the following blog post get released: http://luckytoilet.wordpress.com/2012/12/23/2125/ [...]

[...] http://luckytoilet.wordpress.com/2012/12/23/2125/ [...]

Can anyone tell me, where can I get original WinXP minesweeper? I am playing minesweeper at http://jscr.co/minesweeper/ because it resembles a lot an old one.

[...] LuckyToilet (blog) [...]

[...] +info: LuckyToilet (blog) [...]