Guided Visual Attention Model Based on Interactions Between Top-down and Bottom-up Prediction for Robot Pose Prediction | IEEE Conference Publication | IEEE Xplore