One last feature that we added to our AI was meant to let the AI perform slightly advanced gameplay. The Connect Four game infrastructure is implemented in C and runs on the HPS while the AI logic is implemented in hardware on the FPGA.

Connect 4 Advanced - CNET Download Connect 4 Advanced - CNET Download

However, if we were to parallelize our calculations and span multiple clock cycles, it could be possible to look more moves ahead in the future. We wanted to first implement the overall framework of the game, and then implement more of the AI behavior as we progressed rather than trying to implement recursion and Minimax from the get-go.

At first, we had all our crossCheck wires were red, which indicates that Quartus optimized some bits away.

Connect Four Advanced By MB Games - Classic Toys

The maxConnectFinder modules receive as inputs the rows of the generated game state and use the same heuristic to produce a value corresponding to each of the seven possible moves. As mentioned above, the look-up table is calculated according to the evaluate_window function below. As such, based on the player who has just played, the checkGameOver() function will check for certain patterns (i. In the above image, we see that if the human player messes up and places their token in column 3, the AI will be able to win on its next turn. In fact, the 7 can be shaped in any way: place the disks to the right, back, and even top to bottom.

Each of these correspond to the move that the AI determines is the most likely for the human player to play given each of the possible game states. From the images, we see that ModelSIM expects the AI to output a move in column 3 (represented in binary as 2, since we start indexing at 0).