Feasibility of thesis prototype

Ok fyi I am building an electronic arbiter for the Game of the Generals ( if you are familiar with the game) and its board is made up of 8 rows and 9 columns, so in total there are 72 rectangles. There will be 144 pair of proximity sensors beneath to determine positions. Why 144 if there are only 72 rectangles? It is because each rectangle is divided into two, each half is exclusively for each player. The electronic arbiter will be a robotic arm ( that's why i mentioned earlier that i will be needing 6 motors). The arm will only activate if both proximity sensors beneath a single rectangle senses an object ( that is called a "challenge" in the game). As a response, the arm will determine the position of the challenging pieces and goes to it to judge them. The ranks will be determined by the barcode scanner, and the losing piece will be picked out of the board automatically, leaving the winning piece in that certain rectangle. The proximity sensor module to be used is just the common IR LED and photodiode, resistors, potentiometer, and the LM358 op-amp (i thought there is no need to tell the details of it because it is common proximity sensor). So, back to what i am asking, how many MCP23017 can be connected to a single arduino mega or to a single I2C bus? Sorry if im too persistent and annoying, im just a beginner on arduino thats why im asking the feasibilty of my project to avoid waste of money and time .