Automatically constructing an action knowledge base for an intelligent robot using GA

摘要

The learning algorithm to automatically construct knowledge of action in order to achieve tasks that is given to the intelligent robot is reported. Conventionally, the knowledge of action for a robot was mostly constructed by humans, but the more complex the information of the environment, action, and task becomes, the more trial and error must be repeated. Therefore, the robot must learn and construct the knowledge of action by itself. In this study, the action to achieve a task in an environment is generated by genetic algorithm, and it is shown by experiment, that by repeatedly extracting knowledge of action, the construction of the Action Knowledge Base concerning a task of any situation is possible.