Abstract: In this paper, we present a framework for multi-modal data fusion in the latent embedding space to enhance the robustness of autonomous navigation. Autonomous vehicles often rely on cameras ...