Task planning method based on symbol option and action model self-learning

Abstract

The invention discloses a task planning method based on symbol options and action model self-learning, and the method comprises the steps: mapping high-dimensional image data into a symbol state based on a symbol state mapping module according to priori knowledge and digital image processing; based on an option set module, constructing an option set according to the random action and the symbol state; and based on the action model learning module, learning the action model and updating the option set according to the change of the symbol state to obtain a new action model and a planning target, based on the planner module, solving a planning path according to the new action model and the planning target, and exploring whether the new action model exists according to a planning result. By using the method and the device, the action model, the symbol options and the corresponding relation between the action model and the symbol options can be automatically learned, and planning and training exploration are carried out to obtain an action strategy for solving a problem. The task planning method based on symbol options and action model self-learning can be widely applied to the field of symbol planning.

Description

technical field [0001] The invention relates to the field of symbol planning, in particular to a task planning method based on symbol options and action model self-learning. Background technique [0002] Symbolic planning, also known as intelligent planning, is an important field of artificial intelligence. The main idea is: according to the prior knowledge of human beings, the environmental state and action are abstracted into a symbolic representation, using the initial state, target state and action model represented by the symbol, and through the symbolic planner, a path from the initial state to the The action sequence that the target state is feasible, this action sequence also becomes the plan (plan). There are many kinds of symbolic programming languages, such as PDDL or action language, and each language also has its corresponding planner, such as FASTDOWNWARD, CLINGO. Symbolic planning is highly interpretable and does not require a lot of time to interact with th...

AI Technical Summary

Problems solved by technology

However, due to the need for human prior knowledge to abstract the environment and actions, it is difficult to adapt to complex and changeable high-dimensional environments.

Images