000150249 001__ 150249 000150249 005__ 20251017144629.0 000150249 0247_ $$2doi$$a10.1109/LRA.2022.3142915 000150249 0248_ $$2sideral$$a128415 000150249 037__ $$aART-2022-128415 000150249 041__ $$aeng 000150249 100__ $$aRodriguez-Puigvert, J. 000150249 245__ $$aBayesian Deep Neural Networks for Supervised Learning of Single-View Depth 000150249 260__ $$c2022 000150249 5203_ $$aUncertainty quantification is essential for robotic perception, as overconfident or point estimators can lead to collisions and damages to the environment and the robot. In this letter, we evaluate scalable approaches to uncertainty quantification in single-view supervised depth learning, specifically MC dropout and deep ensembles. For MC dropout, in particular, we explore the effect of the dropout at different levels in the architecture. We show that adding dropout in all layers of the encoder brings better results than other variations found in the literature. This configuration performs similarly to deep ensembles with a much lower memory footprint, which is relevant for applications. Finally, we explore the use of depth uncertainty for pseudo-RGBD ICP and demonstrate its potential to estimate accurate two-view relative motion with the real scale. 2377-3766 © 2022 IEEE. 000150249 540__ $$9info:eu-repo/semantics/closedAccess$$aAll rights reserved$$uhttp://www.europeana.eu/rights/rr-f/ 000150249 590__ $$a5.2$$b2022 000150249 591__ $$aROBOTICS$$b10 / 30 = 0.333$$c2022$$dQ2$$eT2 000150249 592__ $$a1.693$$b2022 000150249 593__ $$aArtificial Intelligence$$c2022$$dQ1 000150249 593__ $$aBiomedical Engineering$$c2022$$dQ1 000150249 593__ $$aComputer Science Applications$$c2022$$dQ1 000150249 593__ $$aMechanical Engineering$$c2022$$dQ1 000150249 593__ $$aControl and Optimization$$c2022$$dQ1 000150249 593__ $$aControl and Systems Engineering$$c2022$$dQ1 000150249 593__ $$aHuman-Computer Interaction$$c2022$$dQ1 000150249 593__ $$aComputer Vision and Pattern Recognition$$c2022$$dQ1 000150249 594__ $$a7.6$$b2022 000150249 655_4 $$ainfo:eu-repo/semantics/article$$vinfo:eu-repo/semantics/publishedVersion 000150249 700__ $$0(orcid)0000-0002-6741-844X$$aMartinez-Cantin, R.$$uUniversidad de Zaragoza 000150249 700__ $$0(orcid)0000-0003-1368-1151$$aCivera Sancho, J.$$uUniversidad de Zaragoza 000150249 7102_ $$15007$$2520$$aUniversidad de Zaragoza$$bDpto. Informát.Ingenie.Sistms.$$cÁrea Ingen.Sistemas y Automát. 000150249 773__ $$g7, 2 (2022), 2565-2572$$pIEEE Robot. autom. let.$$tIEEE Robotics and Automation Letters$$x2377-3766 000150249 8564_ $$s3510363$$uhttps://zaguan.unizar.es/record/150249/files/texto_completo.pdf$$yVersión publicada 000150249 8564_ $$s3156596$$uhttps://zaguan.unizar.es/record/150249/files/texto_completo.jpg?subformat=icon$$xicon$$yVersión publicada 000150249 909CO $$ooai:zaguan.unizar.es:150249$$particulos$$pdriver 000150249 951__ $$a2025-10-17-14:25:40 000150249 980__ $$aARTICLE