The invention provides an event tagging method,
system and device based on multi-
modal perception and a medium, and the method comprises the steps: collecting multi-
modal data, and carrying out the preprocessing of the obtained multi-
modal data; carrying out abstraction
processing on the event, taking a VAD
paragraph as a semantic natural boundary, sampling visual, tactile and environmental
modes according to a
fixed frequency, and aggregating each mode in a preset time window to form a layer-by-layer abstraction mechanism from a
signal feature to a time event and then to a semantic event; performing tagging
processing on the event, wherein a standardized event tag comprises an
event type, a refined tag, confidence, duration, an event initiator, an
event object, an intention and time; and publishing and subscribing events, publishing the tagged events through a
system bus, subscribing event tags through an operation module, and outputting the event tag with the highest priority. According to the method, efficient decision judgment of the emotion
robot can be improved, and reuse and management of the decision module can be realized.