AI-Based Multi Sensor Fusion for Smart Decision Making: A Bi-Functional System for Single Sensor Evaluation in a Classification Task.

Feryel Zoghlami,Germar Schneider,Marika Kaden,Harald Heinrich,Thomas Villmann

doi:10.3390/s21134405

Feryel Zoghlami, Germar Schneider + Show 3 more

Open Access

https://doi.org/10.3390/s21134405

Copy DOI

Abstract

Sensor fusion has gained a great deal of attention in recent years. It is used as an application tool in many different fields, especially the semiconductor, automotive, and medical industries. However, this field of research, regardless of the field of application, still presents different challenges concerning the choice of the sensors to be combined and the fusion architecture to be developed. To decrease application costs and engineering efforts, it is very important to analyze the sensors’ data beforehand once the application target is defined. This pre-analysis is a basic step to establish a working environment with fewer misclassification cases and high safety. One promising approach to do so is to analyze the system using deep neural networks. The disadvantages of this approach are mainly the required huge storage capacity, the big training effort, and that these networks are difficult to interpret. In this paper, we focus on developing a smart and interpretable bi-functional artificial intelligence (AI) system, which has to discriminate the combined data regarding predefined classes. Furthermore, the system can evaluate the single source signals used in the classification task. The evaluation here covers each sensor contribution and robustness. More precisely, we train a smart and interpretable prototype-based neural network, which learns automatically to weight the influence of the sensors for the classification decision. Moreover, the prototype-based classifier is equipped with a reject option to measure classification certainty. To validate our approach’s efficiency, we refer to different industrial sensor fusion applications.

Highlights

IntroductionThe topic is called smart sensors [2], which participate in the improvement of productivity and the increase of the turnovers of many industries
With the introduction of Industry 4.0 level [1], we count the integration of new sensor technologies in combination with artificial intelligence (AI)-based solutions in real-world applications every day.The topic is called smart sensors [2], which participate in the improvement of productivity and the increase of the turnovers of many industries
For problems with less complexity in the classification architecture and especially when we refer to a relatively small data set for training, using a deep neural network (DNN) would be overdone

Summary

Introduction

The topic is called smart sensors [2], which participate in the improvement of productivity and the increase of the turnovers of many industries. These benefits have been confirmed, especially when there is an efficient application of the technologies offered in the market. Sensors can behave differently from one environment to another They may deliver data with different qualities [3], which in some cases confuse the AI model and cause classification failures if the model is not stable enough. One misclassification case is costly, apart from the big effort and high expenses that concern the development of an AI-based system dedicated to solving one classification problem

Methods

Results

Conclusion