The process or technology of determining action steps is an important part of solving problems in artificial intelligence, sometimes called planning. Action plans can generally be divided into single-level action plans, multi-level action plans, scripted action plans and opportunistic action plans.
Extended data:
A single-level action plan, a non-level action plan. When making an action plan, you can make a corresponding problem-solving action sequence for each goal, or you can use problem reduction technology. The main disadvantages of the single-level action plan are that everything is not detailed, there is no priority, and the calculation workload is heavy. Sometimes, some non-critical details will make the plan not smooth. For complex problems, multi-level action plans are usually adopted.
Multi-level action plan is divided into several levels of action plan. The basic idea is similar to structured programming, which follows the principle of top-down and gradual concretization. The plan begins with the abstract outline plan, and at the lowest stage, the plan details that can be directly implemented must be put forward in concrete form.
The early general problem solving program GPS has given birth to the idea of hierarchical action plan. Since 1970s, multilevel action planning technology has been further applied and developed in robot operation design system, mechanic training expert system and genetic engineering experiment design system.
Baidu Encyclopedia-Action Plan