elib
DLR-Header
DLR-Logo -> http://www.dlr.de
DLR Portal Home | Imprint | Privacy Policy | Contact | Deutsch
Fontsize: [-] Text [+]

Introspective Robot Perception using Smoothed Predictions from Bayesian Neural Networks

Feng, Jianxiang and Durner, Maximilian and Marton, Zoltan-Csaba and Balint-Benczedi, Ferenc and Triebel, Rudolph (2019) Introspective Robot Perception using Smoothed Predictions from Bayesian Neural Networks. ISRR 2019, 06-10 Oct 2019, Hanoi, Vietnam.

[img] PDF
1MB

Abstract

This work focuses on improving uncertainty estimation in the field of object classification from RGB images and demonstrates its benefits in two robotic applications. We employ a Bayesian Neural Network (BNN), and evaluate two practical inference techniques to obtain better uncertainty estimates, namely Concrete Dropout (CDP) and Kronecker-factored Laplace Approximation (LAP). We show a performance increase using more reliable uncertainty estimates as unary potentials within a Conditional Random Field (CRF), which is able to incorporate contextual information as well. Furthermore, the obtained uncertainties are exploited to achieve domain adaptation in a semi-supervised manner, which requires less manual efforts of annotating data. We evaluate our approach on two public benchmark datasets that are relevant for robot perception tasks.

Item URL in elib:https://elib.dlr.de/133562/
Document Type:Conference or Workshop Item (Speech)
Title:Introspective Robot Perception using Smoothed Predictions from Bayesian Neural Networks
Authors:
AuthorsInstitution or Email of AuthorsAuthors ORCID iD
Feng, JianxiangJianxiang.Feng (at) dlr.deUNSPECIFIED
Durner, MaximilianMaximilian.Durner (at) dlr.deUNSPECIFIED
Marton, Zoltan-CsabaZoltan.Marton (at) dlr.dehttps://orcid.org/0000-0002-3035-493X
Balint-Benczedi, Ferencbalintbe (at) cs.uni-bremen.deUNSPECIFIED
Triebel, RudolphRudolph.Triebel (at) dlr.deUNSPECIFIED
Date:2019
Refereed publication:Yes
Open Access:Yes
Gold Open Access:No
In SCOPUS:No
In ISI Web of Science:No
Status:Published
Keywords:BNN, CRF, introspective classification
Event Title:ISRR 2019
Event Location:Hanoi, Vietnam
Event Type:international Conference
Event Dates:06-10 Oct 2019
Organizer:IFRR (International Foundation of Robotics Research)
HGF - Research field:Aeronautics, Space and Transport
HGF - Program:Space
HGF - Program Themes:Space Technology
DLR - Research area:Raumfahrt
DLR - Program:R SY - Technik für Raumfahrtsysteme
DLR - Research theme (Project):R - Vorhaben Multisensorielle Weltmodellierung
Location: Oberpfaffenhofen
Institutes and Institutions:Institute of Robotics and Mechatronics (since 2013) > Perception and Cognition
Deposited By: Feng, Jianxiang
Deposited On:16 Jan 2020 08:47
Last Modified:16 Jan 2020 08:47

Repository Staff Only: item control page

Browse
Search
Help & Contact
Information
electronic library is running on EPrints 3.3.12
Copyright © 2008-2017 German Aerospace Center (DLR). All rights reserved.