Ref docs : https://forum.pollen-robotics.com/t/how-to-make-your-reachy-plays-tic-tac-toe/15

How did we build the Reachy tic tac toe game?

This demonstration emphasises several key aspects of the robot:

• It involves complex motions that have been defined, saved and can be re-played (grasping a pawn, place it on a pre-defined location).
• It uses the camera, vision and machine learning for recognizing the board and analyzing its configuration.
• It also relies on a simple machine learning algorithm for correctly playing the game itself (what next move should I make).

We divided the game into 4 different sections : Building the hardware setup, Working on computer vision, setting up the program, final integration and testing.

Building the hardware setup:

We divided into the following tasks :

1. Define the game specifications: Board specification, game pieces specification.

2. Design/fabricate board & pieces

Wood was used for the board, game pieces were 3D printed using 3D models and stored in GrabCAD

Complete tutorial of how to create CAD models for Reachy can be found here:

How to create CAD models of 3D X's and O's for a tic-tac-toe game

3. Verified that we do not need a neck angle to be able to see the board.

Working on Computer Vision: