The invention provides a safety belt control method and device, a vehicle and a storage medium, and the method comprises the steps: obtaining image information of a target passenger of the vehicle, seat pressure information and state information of the vehicle, carrying out the
feature extraction of the image information, and obtaining the posture features of the target passenger, performing
feature extraction on the seat pressure information to obtain a
body weight feature of the target passenger, performing
feature fusion on the posture feature and the
body weight feature and state information of the vehicle to generate a
state space of the target passenger feature, performing
reinforcement learning on the
state space by adopting a pre-trained learning model, outputting an action space corresponding to the target passenger, and according to the pre-tightening force and the pre-tightening amount in the action space, the safety belt is controlled to execute pre-tightening operation. According to the method,
visual detection,
pressure detection and multi-
modal data fusion of the state information of the vehicle are utilized, accurate recognition of the characteristics of the passengers in the vehicle is achieved,
dynamic mapping of the characteristics of the passengers and the pre-tightening parameters is established, and a personalized safety belt active pre-tightening constraint strategy is provided for different passengers.