| abstract |
The invention discloses an end-to-end semantic instant positioning and mapping method based on deep learning. The continuous original image sequence and the original 3D point cloud sequence corresponding to the image sequence are respectively collected by the color camera and the lidar, and the pose transformation information, depth information and semantic segmentation information of the continuous five-frame image sequence are processed; The multi-task deep neural network is input into the multi-task deep neural network, the multi-task deep neural network is trained to obtain parameters, and the trained multi-task deep neural network is used to process the image sequence of five consecutive frames to be tested, and the interval between image frames is obtained. pose transformation information, depth information and semantic segmentation information. Compared with the traditional ORB-SLAM algorithm and the method based on deep learning, the method of the present invention has better performance. |