Introspective Robot Perception using Smoothed Predictions from Bayesian Neural Networks
This work addresses uncertainty estimation for robotic perception, offering incremental improvements in accuracy and reduced annotation effort.
The paper tackled the problem of improving uncertainty estimation in object classification from RGB images for robotic applications, achieving performance gains by integrating Bayesian Neural Networks with Conditional Random Fields and enabling semi-supervised domain adaptation.
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 (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 in annotating data. We evaluate our approach on two public benchmark datasets that are relevant for robot perception tasks.