Multi agent pathfinding
Multi agent pathfinding (or MAPF) is an algorithmic problem where agents - virtual robots - move around a grid or graph to a goal location. The algorithm plans their paths while avoiding collisions. Finding optimal paths in which the total travel time is minimized is an NP-Hard problem.
This website is now in it's second generation. In the first year, a team of students tried to adapt existing MAPF algorithms to allow for waypoints to be added in the agent's routes. Results of this research are still available here.
In this second iteration, existing algorithms are again extended to allow for so-called "matching". With matching, Agents don't have one goal location, but instead a number of them, shared with other agents of the same color. Together they need to find an optimal path to their goals all the while avoiding collisions with agents of any color.