The robotics 2 team I was a part of (I’m in the middle) created a Pacman implementation with Lego robots. Unfortunately the ghosts couldn’t literally eat Pacman (The university didn’t want its lego robots to be damaged during development), instead they had to trap Pacman. If pacman was cornered on any part of the playing field, the ghosts would win, if Pacman ate all of its food Pacman would win! As a bonus we implemented the possibility for people to control Pacman, so they could try and beat our ghost AI’s.
Prize winning team
The robotics 1 team I was a part of won the best robot of the Univeristy award, it got the highest score of all contestants. The robotics 2 team got the highest grade (along with a few other groups) and came in second for the public’s favorite prize.
The implementation relied on the Lego brick framework and Matlab. The robots could not communicate with each other, but Pacman had to evade ghosts and the ghosts had to trap Pacman. To achieve the multi-agent cooperation we used Matlab and a Bluetooth connection. The robots would pass on their location to Matlab and Matlab would use a utility function for each specific robot to determine the next action to take. This meant our implementation required both searching and planning, in the end it all worked!