Python Checkers AI Tutorial Part 2 - Implementation & Visualization (Minimax)

[@thomasbezencon2121]

Hey Tim, great video. I think the reason why the AI king kept boing back and forth on the last two rows, is because of a bug in your program. If a king lands on the "king row", the king counter is increased despite no new king having been created (a piece can't be a double king). Given that this king counter is part of the algorithms scoring function, it used this bug to increase its own perceived score.

[@aaronwtr1150]

I noticed this as well, and I made sure that a single piece could not become a king more than once and even then this issue persists... @techwithtim any suggestions?

[@aaronwtr1150]

I have been playing around with it for a bit now and I think I have narrowed down the problem. The minimax seems to consider a single piece at a time. If the piece has valid moves available, it takes them. If there are none, it skips to a next piece. So when a king is made and this piece it selected, the king will just move around, since its moves are valid. I am to find a way to fix this ..

[@MD-ji7dh]

@@aaronwtr1150 were you able to solve the issue? Can't find any solutions and it's pretty anoying for the game to end after on ai got one king

[@aaronwtr1150]

@@MD-ji7dh I have fixed that particular issue, but there still are some bugs even if that bug is fixed. For example, the AI seems to select a single piece at random and ONLY moves that piece until it is captured or makes a king.

[@MD-ji7dh]

@@aaronwtr1150 could you please tell me what's the solution of the repeating king? I have tried for so long to find the issue

[@spark1539]

just want to say your a god send, got a project for AI class and the teacher showed us no code on how an ai might look or function so this was and is endlessly helpful.

[@bradmclean4988]

Sounds like we have the same class

[@justins7796]

same here. Is your teacher's last name Li?

[@bruhmomentum0734]

same here

[@softknk1422]

The idea of you with storing possible moves in a map data structure helped me a lot .. Thanks for that!

[@Fersko]

It's great to have someone who speaks normal English and can explain the programming properly. Thanks a bunch, Tim

[@dpskane]

I deviated with my coding during the first 3 parts and I am glad that my board state has a list of the pieces instead of the actual number. Makes the whole process of finding pieces easier.

[@DarkusNights]

Old tutorial but this actually helped me for my classes and adapt your methods to my own code.

[@chinonsodimgba7689]

Nice one, I followed this tutorial and have been able to implement the checkers game using dart and flutter, but the pain comes in the AI keeps making unimaginable wrong moves, mostly because they do not know that it is possible to kill more than one enemy in one turn and the position of the pieces do not matter, so for that reason I have been working on making the algorithm take note that a player can kill more than one enemy in one turn and improve on the scoring formula

[@amankaushik5833]

Hi tim thanks for making the second part! as always I just love watching your videos related to any topic!

[@shaharyarahmed5777]

Recursion is always confusing but after a lot of thinking you can finally get your head around it.

[@manfredreinsch8448]

Hi Tim, very nice tutorial, well done! I have one question. one remark and two ideas how to extend and even improve your checkers tutorial. In your evaluate method there is a factor of 0.5 for the king piece. Because a king piece is more valuable than a normal piece, shoudn´t that be a factor of 1.5 ? Of course your algorithm will find the best way to win or just to avoid a lose according to your evaluation method , but it can not handle a draw. Consider an obviously drawish situation, e.g. just one red and one white king remaining; the AI will keep on trying to win, while the human player gets bored and sacrifices his king to end the game. In chess there are rules to avoid such a situation. You might add a chesslike rule for offering, accepting or even force a draw, when there was no capture amoung the last x moves. Yes, I am aware that you considerd alpha-beta pruning and rejected it for your current implementation, probably because it seems to be too "technical" for you. Still alpha-beta pruning will accelarate your algorithim by a factor of one ore even two according to depth. Therefore you can easily show, how a slight change in your code will improve an algorithm by a great amount. Because you have implemented a representation to show the effects of the speed of your algorithm depending of the depth in a very nice way, you can present the accelation by alpha-beta pruning visualy. If there are only a few pieces left on the board, you might consider to add an end-game database in your program, which is i.e. offered by the Canadian world best checkers program Chinook for free use. This might be even an idea of merging your checkers- and Mango-DB-Tuts...

[@fisheatsyourhead]

There is meant to be a 25 move rule where 25 king moves each with no captures forces a draw

[@victor20214]

Tim has been an inspiration to code and teach code. I'm from a country where coding isn't as popular and learning it through time has been great. and maybe its time for the bird to fly on its own. hope you will help my endeavour in this quest

[@Kobbiemainoocomps]

i know what you mean. ill support you.

[@williamikennanwosu]

Thanks for the video Tim!

[@ashwanirai8324]

Sir, if I am doing it for AI vs AI it is showing error at the end states that Attributeerror Object has no attribute winner

[@anzarrabbani3766]

36:43 can some1 pls help me, it keeps saying: File ---------------------------------------------------------------------, line 43, in get_all_moves for piece in board.get_all_pieces(color): TypeError: 'int' object is not iterable

[@ohgreatone1381]

Tim great video but there are several "bugs' First as you and anther person mentioned the white king goes to the bottom and stays there. Bouncing back and forth between two squares. I believe someone found a solution but it did not get posted. Additionally, randomly the White King will move from the lower square where it is bouncing around and suddenly goes 6 squares diagonally to jump over a red piece. This is not supposed to happen in any version of checkers I have ever played. Could you please post the correction for the White King bouncing around. Thank you

[@showmickkar7793]

Hey Tim, just out of curiosity, in a video you said you do all you programming on linux. Why do use windows for making tutorials?

[@ahmedthegreat3973]

He Said That Most Good Video Editing Softwares Are On Windows

[@КириллПопов-о3с]

Hey Tim! My english is bad, but i try to ask you a question. In your minimax function you do something like that for move in get_all_moves(position, WHITE, game): . So you store all game tree. Can you answer how much memory your program uses with depth different depth. Ty!

[@gilbertocisneros85]

Can this be implemented with the real checkers rules? The jump in not optional and with different levels? In a website

[@ohgreatone1381]

Also, you mentioned that you were going to show us how to make the computer play itself. I tried adding a if game.turn == RED statement like the WHITE statement but it only allows the white pieces to move and the game ends with the white pieces moving down and blocking the red pieces.

[@manfredreinsch8448]

Hi Tim, I have just found another minor problem: It is possible to blockade your opponent, i.e. he is unable to draw while you can make one or more moves. So "no move" shoud be an option for possible moves!

[@fisheatsyourhead]

no possible moves means you lose, even if you have pieces on the board. Unlike in chess where no possible moves with pieces is a draw.

[@sachinsinghal13]

Hey Tim, Thanks for everything, the other checkers series and this checkers series were awesome. Though, I have been trying for long time to make the game so we can undomoves as well, but I'm unable to. How can we make the unndomove function?

[@sevendoesstuff]

if you save all the moves played in a list then when the back arrow (or whatever key you want) is clicked delete the last move done and then quickly play all the moves in the list from the start position then do display.update() that should do it. you could also save all the board positions in a list / matrix and delete the last position in the list and set it to the now last one (the one before it). the ladder takes up much more space as you have to save every single board position from the start while the first just saves 2 points (where it was and where it moved too) but it is a bit more complex.

[@AliHaider-bc7cl]

i keep on gettin the game object not iterable error.. any help?!

[@matthewzamat3331]

Hey Tim! Thanks for the video. Is there any way we could create a function to make this checkers AI play against itself? how would we have to modify this code?

[@ohgreatone1381]

Also found that the list of valid moves for RED fail when approaching row 0. I can double jump everywhere on the screen except when the last move results in a KING. White can do a double jump but not the RED player

[@crjcpj9100]

Hi Tim ,i really like your color theme on vscode how did you do it. what's it called.

[@tomiolah5137]

I haven't watched this video yet but I plan on doing it soon. Is it possible to build a chess AI based on this?

[@wiwikkarlina4447]

thanks for the video! but i still dont get how the static evaluation of position in minmax at depth = 0 get passed up to the parents?

[@MD-ji7dh]

if you debug you can see the return statement isn't exiting the whole function but the current recursion

[@AfrasiabAhmadSU__O_OO_OO]

which color theme and font are you using currently

[@SnoozeTheRecluse]

I keep getting AttributeError: 'NoneType' object has no attribute 'winner'

[@divyanshsingh2614]

Hey bro, can u please bring a tutorial on React.Js It would be of great help.

Hey Tim! Is your website down or you are using a new domain?

[@programmatore]

where can i begin this game project....new programmer ho can understand this..............................000000000000 value

[@ruchi727]

you got a new sub!!!

[@EmmnauelPeace]

Thanks so much for this hard work this can further be use in robots to make them take better decision as humans ...Well we need to explore more on this min-max algorithm with neural networks and others .When these algorithms are used in combination it really makes it very difficult for humans to beat computers in context. Great work..Best regards

[@tiagozpt7743]

Friend u know gamezer ? Rules u cant create this rules checkers ?

[@ahmedthegreat3973]

Tim I Cant Access Your Website Its Shows Account Suspended. If Its A Bug Please Fix It Please

[@richardtarbell946]

I also ran into this issue earlier. I think it is fixed now.

[@ahmedthegreat3973]

@@richardtarbell946 Okay Ill Check Thanks For Letting Me Know

[@ahmedthegreat3973]

@@richardtarbell946 No I Checked It. Its Not Fixed

[@richardtarbell946]

@@ahmedthegreat3973 You're right my b. I was looking at this link tech-with-tim.teachable.com/ rn other one is still down.

[@namo2943]

nice i just open youtube and saw this

[@alcalina636]

Hello, how are you? did your site just get down? tks

[@kae4881]

nice video!!

[@kotvkvante22]

where can I donate to you? paypal doesnt work for me

[@avigupta6064]

Donate me instead😔

[@kotvkvante22]

@@avigupta6064 did not deserve

[@techwithomar1875]

I was literally watching the checkers video

[@RohanKumar-zd3pc]

Hey Tim, When I tried to reach techwithtim. Net then it showed "account suspended"

[@shaharyarahmed5777]

Same here

[@danielros7798]

I got this error and don't understand how to fix it File "main.py", line 47, in main() File "main.py", line 27, in main value, new_board = minimax(game.get_board(), 4, WHITE, game) File "C:\Users osda\Documents\Python\Checkers met AI (MiniMax)\minimax\algorithm.py", line 15, in minimax evaluation = minimax(move, depth-1, False, game)[0] File "C:\Users osda\Documents\Python\Checkers met AI (MiniMax)\minimax\algorithm.py", line 8, in minimax if depth == 0 or position.winner() != None: AttributeError: 'list' object has no attribute 'winner'

[@johnathanindigo3340]

same here, did you manage to fix it?

[@danielros7798]

@@johnathanindigo3340 that's a lot of code, I am going to my PC now and then I'll send it :)

[@danielros7798]

@@johnathanindigo3340 hey, I don't know exactly what I changed so I have the whole code here in a hastebin: hastebin.com/adoqinidom.properties

[@johnathanindigo3340]

@@danielros7798 hey, so this one works?

[@danielros7798]

@@johnathanindigo3340 Yes

[@markgatere5007]

Hello Sir, Am a High School student. Kindly advise me and other high school guys who have interest in tech on Computer Science majors in University or college to later have an internship or job in the top high-tech companies like Google.

[@TheFonurb]

After following this tutorial, i realized i'm probably the worst checkers player in the world... even the 1 depth AI could beat me one time T.T

[@artmancs4108]

I'm doing a project where we want a robot playing as the AI, for this I need to have the coördinates (row, col) of the final move it took in a list variable. That way I can link these variables to points for the robot. I've been struggling to get this variable out of the program. Should be simple but I can't figure it out haha. Someone help this poor fellah out

[@tahalyousfi]

Great work Tim

[@renuaggarwal7660]

hey bro pls. provide link to the Asset file too . it would be a great help

[@shiehuapiaopiao]

It's in the description.

[@georgianabhishek3561]

perfect just opened pycharm

[@duncansutherland5917]

goat

[@gilbertocisneros85]

This is not American checkers rules because jumps must be mandatory

[@ganapathypt6410]

🖤

[@ahmedthegreat3973]

Please Create A Video About Creating A Slack Bot In Python

[@MD-ji7dh]

pretty sure he has one

[@נועםאיפרגן-ל7ע]

44:40

[@PAuLXPlain]

Assistant Didn't work Please help

[@alenvarghesec.k8001]

Why not make python hacking series? (Or you can't do dat)

[@WForWonder]

kather.......................................

[@MD-ji7dh]

Not bad, but a lot of bugs

[@techwithomar1875]

Hi

[@baburajp8342]

Tenth comment 😉

[@karansingh9700]

second

[@dogesh828]

third

[@prajwalmishra8562]

First

[@tahalyousfi]

Hi I love your content Tim