The line following problem is some kind of introduction task for robot control. The goal is program a physical robot to follow a black line on the ground. What the programmers have to do is to handle the steering control. In it's easiest case this is equal to implement a pid controller but some advanced strategies like behavior trees and neural networks are utilized also. if the line was interrupted or the robot is bit away from the course then the single robot is still able to follow the line, .
Asking how a line following robot has to be programmed would be too easy, because lot's of tutorials and working code is available already. Github for example provides an endless amount of repositories in which the problem is explained, solved and even realized on a concrete hardware architecture, for example an ARM microcontroller.
It make sense to raise the abstraction level a bit and ask how a line following strategy can be used in a broader context. The line on the ground is equal to a model. It describes what the robot should do on a map. And the robot is tracking that model by following the advice of the line. If the line goes to the left, the robot has to do the same. What will happen if the number of robots on the map is increased to 10 and more? Right, then the scene will look similar to the lemmings game. That means a swarm of AI artificial life agents is following a plan on the ground, and all of them will reach the endpoint of that line.
To make things a bit shorter, the idea is to solve the Lemmings game (which is an np hard AI problem) with a line following algorithm in which a swarm follows a path on the map. All what the Lemmings have to do is to obey the given trajectory. The actions of the Lemmings have to make sure, that they are following the line, similar to the situation for a single robot.
Does this description make sense, or is solving the Lemmings game a bit more complicated?