scheduler (Intro to Artificial Intelligence Project)

Developed a solution to a scheduler problem in a group of 3 people.

Problem

Given a schedule and a number of tasks, the program schedules when each task should be completed, taking into account the properties of the task, which include:

• the task's priority level
• the task's due date
• the task's estimated time span needed to complete
• if the task can be broken up into subtasks
• the number of activities that can be multitasked with the task
• a list of tasks that need to be completed (prerequisites) before this task

Solution

scheduler_csp.py

To solve this problem, we constructed 2 classes: Task and Slot.

Task

This class contains all the attributes associated with the task, as listed in the problem section. This class has a method split(), which is called on a Task object to try split the task into subtasks. It returns the list of subtasks, or a list containing itself if it cannot be split.

Slot

This class represents a timeslot in the schedule, with an attribute -assigned, a list of tasks that have been assigned to this timeslot- and 4 methods. These methods include:

• assign(self, task)
• unassign(self, task)
• get_capacity(self, candidate=None) If the Slot has not been assigned a Task, the slot is available, "inf" is returned. If there a Task has been assigned to the Slot, the value returned will be the smallest multitask value, indicating if more Tasks can be assigned to this Slot.
• next(self, slots) Returns the next timeslot in the schedule.

propagators.py

Implemented 3 propagators:

• prop_BT backtracking propagation
• prop_FC forward checking
• prop_GAC GAC propagation

cspbase.py (Provided-I did not write this file)

Provided classes for CSP, containing classes for: Variable, Constraints, CSP and BT

scheduler_testing.py

A small testing suite to test program functionality