MAAL: Multimodality-Aware Autoencoder-based Affordance Learning for 3D Articulated Objects | IEEE Conference Publication | IEEE Xplore