Table 3.

Challenges and countermeasures of metalearning.

	Technical aspects	Application aspects
Specific problem	Costly Poor learning ability for complex tasks Single learning method of metaknowledge	Difficulty in obtaining data Inconsistent task distribution
Response plan	Improving model training performance Studying the continuity of action execution Metaknowledge combination learning	Building a training task database Learning from the experience of other algorithms [67–69]