Robot emotion cognition method based on depth learning

Abstract

The invention provides a method for robot emotion cognition based on depth learning, which comprises the following steps: obtaining multimodal information, and obtaining emotional state through a multimodal information fusion algorithm; obtaining emotional state through the multimodal information fusion algorithm; inputting emotional state into an interactive decision model for processing to obtain decision information, and outputting the decision information to a satisfaction degree model; satisfaction degree model being used to detect whether the decision information outputted from interactive decision model accords with the current user's preference. If so, it enters into long-term memory, updates the data, and outputs the obtained decision information to behavior selector to obtain robot behavior output. Otherwise, the recorded fragments and the corresponding emotional changes of the user are retrieved from the short-term memory, the user characteristics are extracted from the long-term memory, the decision information is calculated by using the LSTM depth learning model, and the obtained decision information is output to the behavior selector to obtain the robot behavior output. The invention improves the cognitive ability of the robot and better performs human-computer interaction through an interactive decision model.

Description

technical field [0001] The invention relates to the field of information technology, in particular to a method for robot emotion cognition based on deep learning. Background technique [0002] Judging from the current achievements in robot emotional cognition, calculations are mainly based on physical behavior, and there is no progress in emotional cognition. The existing methods have the following defects: 1. Most of the robots on the market belong to the category of industrial robots, and further human-computer interaction based on human emotion is not considered in the interaction process; 2. All robots have decision-making modules, but none of them are based on emotion And making behavioral decisions is not conducive to the development of social behavior between robots and humans; 3. Existing research rarely has an emotional cognition framework based on deep learning, and cannot interact with different objects, nor can it perform better self-awareness renew. Contents ...

AI Technical Summary

Problems solved by technology

The existing methods have the following defects: 1. Most of the robots on the market belong to the category of industrial robots, and further human-computer interaction based on human emotion is not considered in the interaction process; 2. All robots have decision-making modules, but none of them are based on emotion And making behavioral decisions is not conducive to the development of social behavior between robots and humans; 3. Existing research rarely has an emotional cognition framework based on deep learning, and cannot interact with different objects, nor can it perform better self-awareness renew

Images