The invention provides a memo generation method. The method comprises the following steps of receiving multimodal input information from a user, wherein the multimodal input information at least comprises
voice data; analyzing the multimodal input information, and extracting the
voice data; converting voice generated by the user into text data; performing semantic comprehension on the text data, and capturing a keyword of a function of entering a memo from an intention result of the semantic comprehension; and enabling a memo item recording function, and outputting memo recording success information. In a convenient and quick input mode, the intelligent
robot assists the user to memorize big and small events. When a query needs to be performed, the intelligent
robot is enabled to accurately know a user intention through accurate semantic comprehension, thereby giving out a corresponding answer.