Wifi for Arduino ESP32

[rangoy]

I bought the Arduino ESP32 to use with the board.

It seems like the WIFI-library used in the project does not support the WIFI in the ESP32 module.

Is support for the wifi module in ESP32 planned?

[Chris0xdeadbeef]

Yes ESP32 is used for WIFI, but you need to get the library check it on https://randomnerdtutorials.com/esp32-useful-wi-fi-functions-arduino/

[joojoooo]

WiFi Module: For AI gameplay (built-in on RP2040 Connect)

I too wanted to use an ESP32 I already have but the README states the WiFi is only for the Arduino Connect so I'm not sure how to compile for the ESP32 with the WiFi functions working too

[rangoy]

I've started an effort to add support for ESP32.

I have a running version at this fork; https://github.com/rangoy/Open-Chess.
Caution: it's vibe coded and it has to be reviewed it througly before It might can be pulled back

[joojoooo]

it's vibe coded

😢 Did you test it at least? Does it work? did you quickly read the generated code to see if it makes any sense or did you just copy paste whatever?

I plan to do an esp32 fork too, for my plastic "PCB" but it will have different pins used and different LED/sensor layout. The WiFi code should be the same

[rangoy]

The current code works both, starts an AP and also connects to a configured wifi. I vibed further to get stockfish integration and some other features. I did skim through the code but it's been years since I've coded Arduino.

I (or my friend Cursor) have added and tested:

• Connect to AP using ESP32
• Display board in browser to check that the state is correct
• Edit board. (Somewhat works, blocked when a piece is picked up)
• Set colors during placement of pieces to indicate direction of the board (white piceses on white leds, black on blue leds. Red means you've placed it wrong
• Connects to stockfish
• Color LEDS to indicate game selection so you only have to remember color and not a placement

Everything seems to work, but webserver blocks while wating for a move

Not tested: other hardware, using WIFI-config in web-interface

The board is not very nicely styled, the black pawns does not have the same look as the other pieces. The edit interface is also a bit ugly.

So as far as I can see it's working fine, but as I usual the AI is not lazy enough to provide compact and clean code but rather verbose and somewhat suboptimal

[rangoy]

It seems like there was a mismatch between row/col mappings, I'll try to sort it out

[joojoooo]

@rangoy Your AI code really sucks 😆, I'm painfully improving it on my fork, with the help of even more AI, but I review and edit it so it's good. My strip layout is a bit different so if you want to use it you will have to change the GPIOs and pixel index mapping for it to work, it's really easy to change. I'm also using the cheaper LEDs without dedicated white, just change led type back to GRBW and it will work for yours.

[rangoy]

@rangoy Your AI code really sucks 😆, I'm painfully improving it on my fork, with the help of even more AI, but I review and edit it so it's good. My strip layout is a bit different so if you want to use it you will have to change the GPIOs and pixel index mapping for it to work, it's really easy to change. I'm also using the cheaper LEDs without dedicated white, just change led type back to GRBW and it will work for yours.

Yeah that's true! I started to restructure that in a branch https://github.com/rangoy/Open-Chess/commits/restructure/ - but I lost focus.

I tried to build the frontand as a SPA instead of these all these crazy concatenated strings.

Exciting. I'll check out your fork later.

[joojoooo]

@rangoy I already fixed about all the main issues:

• Switched IDE from ArduinoIDE to Visual studio code platformio. (the AI there did all the following fixes, I just chatted A LOT, verified the output, chatted more, and finally tested if it actually worked...)
• Implemented castling, check w/animation (blink), checkmate, stealmate detection, game over w/animation (fireworks)
• Queen and King were swapped, black and white sides swapped.
• Human vs Human now enforces turns !
• Now chess_moves and chess_bot have the same colors behavior
• 100ms sensor denounce at the board_driver level. Allows you to slide the paws forward 2 squares without getting them detected on the first one. (if you're fast)
• Crazy mess web page concatenation fixed with a really elegant platformio phyton script that runs on build and automatically takes all the pages, minifies and compresses them, then generates the cpp files with PROGMEM arrays of the pages. Also generates cpp for an automatic page router. All that means you can simply add an html/css/js file to the web folder and it gets served automatically
• Blocking web server replaced by the Async one, so the main loop can't block board updates / client requests
• Immediately update webUI board state as soon as it changes, you no longer need to wait for the bot turn to end to see the board update in the web UI.
• Rewrote stockfish json parsing mess/AI hallucinating inexistent fields, now it properly parses all real fields into a struct with a proper json library
• Many web UI fixes: new working evaluation bar, easier piece selection on board edit, json board updates (no crazy long html with all the squares and whole page reloads). Working WiFi credentials saving form that actually connects and saves them, even a reflash doesn't delete them. Removed useless "future" form fields. Made it more pretty with new buttons, less useless text, etc..
• Now Bot mode doesn't disconnect and re-connect to wifi every time it's started if the ESP32 is already connected
• New WiFi connecting animation, showing all the attempts instead of static lights

I was flabbergasted when I saw the original source code being SO bad. All the people who bought a PCB... how do they even use it? by default it's impossible to finish a game against a bot due to the lacking castling/check the bot will just get stuck "thinking" forever. And chess moves mode is not much better, it suggests illegal moves, doesn't enforce turns, doesn't let you castle, king and queen starting in the wrong squares, board orientation is wrong, (white not on the right) ...

[joojoooo]

@rangoy I just added automatic calibration , it will automatically rotate your board and map all the LEDs to the correct square. No matter what order of GPIOs is used, no matter what order the shift register output uses, no matter what layout the LED strip has. The board has the wrong orientation with the original source code and PCB (white bottom corner square not on the right). Since my PCB is built manually, it's really easy to mess up orientation or pin order, so that's a really cool feature. So if anyone wants to test it, simply change the pins here and flash

[rangoy]

Nice @joojoooo ,

I think I'll have some time next week to check it out.

I also realized that the mapping was wrong, and I was thining it was an AI error / mapping error when implementing the support for ESP32.

I was also disappointed pretty in the status. The Print & Play were rather Print & Pray [that someone will implement it].
I was thinking that I could upload something playable and tune it.

I'm also surprised that they squashed the intial commit of the repo, is someone trying to hide licensing issues? There's a fork from before the only commit here; https://github.com/sphilius/Open-Chess/commits/main/

Do you have any idea on how much has to be backported to use this on the "original" board?

[joojoooo]

I just finished adding ALL the chess rules, en passant too, now it's fully playable. I completely re-made the website too with a new chessboard, piece drag and drop features, FEN copy/edit, en passant square highlight, SVG pieces instead of text, the board also shows ranks and files.

[Gobux]

Can someone help me,
I am using the ESP32 also as i didnt have the RP2040.

I can only get some LED's to light up but it only activates on 1 row with the magnet., the other rows are not responding... i dont know why..

[joojoooo]

@Gobux I can try to help you, flash my fork. On first boot all leds light up one by one, this will tell you if the led strip is working correctly. Then calibration will start and will print all the necessary info about GPIO/SR pins in the serial monitor. You can easily spot any weird pattern using that info.

The issue is for sure the original PCB being really bad, it can't realibly read the sensors, it's an electrical design issue, I already helped someone else with the same exact issue. You need to cut a trace and add a level shifter and transistors, only then it will work reliably... more info on my repo and on my MakerWorld

If you have more questions or need more help, message me privately, let's not spam here, it's unrelated