Automatic parking method and system based on multi-sensor information fusion

The invention relates to an automatic parking method and system based on multi-sensor information fusion. Specifically, the automatic parking method comprises the steps that deep learning parking space detection, traditional visual parking space detection and ultrasonic parking space detection are carried out based on 360-degree look-around images and ultrasonic radar data through a parking space detection fusion algorithm, and detection results of the deep learning parking space detection, the traditional visual parking space detection and the ultrasonic parking space detection are fused; the vehicle is positioned in real time through a visual positioning algorithm; detecting obstacles around the vehicle through an obstacle detection algorithm; obtaining a full-information map through an electronic map fusion algorithm; and through a parking planning control algorithm, a parking track is planned, and the vehicle is controlled to complete parking. 本申请涉及基于多传感器信息融合的自动泊车方法及系统。具体而言，该自动泊车方法包括：通过车位检测融合算法，基于360度环视图像