Planning, scheduling and optimization
It helps decide what to do, in what order and with which resources when alternatives compete. A sound formulation makes the goal and constraints explicit, such as capacity, time windows, distance or safety, before calculating routes or schedules.
Text reviewed on
← Back to the 7 pillars
When it emerged
Planning actions and choosing good routes predate modern AI. Between 1966 and 1972, the Shakey robot used the STRIPS planner to break goals into possible actions, a milestone in AI planning research. Optimization also inherited methods from operations research and applied mathematics.
What it is
Planning selects action sequences to meet goals; scheduling allocates tasks over time and resources; optimization compares alternatives under objectives and constraints. Search, heuristics and evolutionary algorithms are among the possible techniques.
How it works
A problem is described by an initial state, a goal, possible actions and constraints. A planner searches for a feasible sequence; a scheduler allocates tasks over time; an optimizer compares solutions by cost, time or another objective. Exact algorithms may seek a proven best answer, while heuristics find a useful one faster.
What it is for
It helps decide what to do, in what order and with which resources when alternatives compete. A sound formulation makes the goal and constraints explicit, such as capacity, time windows, distance or safety, before calculating routes or schedules.
Where it is used
Deliveries, work schedules, machine allocation and robot paths are typical uses. In delivery routing, each vehicle has a capacity, each stop may have a time window, and the system seeks a feasible combination with lower estimated cost.
Limits and safeguards
A computed solution depends on the data and constraints; it is not necessarily the best possible and does not require predicting the future. A computed solution reflects the data and priorities supplied. For large problems, finding a provably optimal answer may be impractical; routing tools may return a good solution that is not the best possible. Traffic or demand changes require recalculation.