END-TO-END-DEEP-GENERATIVE MODEL FOR SIMULTANEOUS LOCALIZATION AND IMAGE

摘要

This disclosure provides an end-to-end deep generative model for simultaneous localization and mapping. The disclosure relates to systems, methods, and devices for simultaneously locating and imaging a robot in an environment using a generative adversarial network with a varying auto-encoder (VAE-GAN). One method includes receiving an image from a camera of a vehicle and providing the image to a UAE GAN. The method includes receiving reconstructed pose vector data and a reconstructed depth map based on the image from the UAE GAN. The method includes calculating the simultaneous location and mapping for the vehicle based on the reconstructed pose vector data and the reconstructed depth map. The method is such that the UAE GAN comprises a single latent space for receiving a variety of inputs.