Sensor Fusion Algorithms for Robotics: Bayesian Inference vs. Cortical Circuits

نویسندگان

  • Florian Bergner
  • Martin Buss
چکیده

Sensor fusion is a commonly used technique to fuse information from multiple information sources in such a way that synergistic effects are used and the result gets more reliable. Sensor fusion is needed by robotic systems which have to navigate safely and autonomously in a noisy and dynamic environment. The robot’s system state in the environment is usually estimated by mathematical and stochastical based fusion techniques like Bayesian Inference, Kalman Filter, etc. Recent breakthroughs in psycho-physics, neurophysiology and bio-inspired computing have brought forth new sensor fusion techniques like Divisive Normalization. Divisive Normalization enables sensor fusion at neural network level and fulfills many neurally plausible empirical principles which conventional mathematically based fusion techniques cannot show. This work explains and compares Bayesian Inference and Divisive Normalization and exemplarily shows their implementation on robotic systems. The advantages and disadvantages of these implementations are discussed afterwards. Zusammenfassung Sensor Fusion ist ein etabliertes Verfahren um Informationen verschiedener Informationsquellen so zu kombinieren, dass Synergieeffekte genutzt werden und somit zuverlässigere Schlussfolgerungen gezogen werden können. Sensor Fusion wird von Roboter-Systemen verwendet, um in einer sich ändernden Umgebung und bei verrauschten Signalen sicher und autonom navigieren zu können. Der Systemzustand eines Roboter-Systems in seiner Umgebung wird üblicherweise mathematischstochastisch geschätzt, wobei Fusionsverfahren wie Baysian Inference, Kalman-Filter und viele weitere angewendet werden. In jüngster Vergangenheit hat es einige Fortschritte in der Forschung geben, vor allem in den Bereichen Innere Psychophysik, Neuro-Physiologie und Biologisch-Inspiriertes-Computing. Diese Fortschritte haben die Entwicklung neuer Fusionsverfahren angestoßen wie z.B. Divisive Normalization. Divisive Normalization ermöglicht Sensor Fusion auf der Ebene von neuronalen Netzwerken und steht im Einklang mit vielen plausiblen, biologisch-empirischen Prinzipen. Viele mathematisch basierte Fusionsverfahren hingegen erfüllen diese Prinzipen nicht oder widersprechen ihnen sogar. Diese Arbeit erläutert und vergleicht die Fusionsverfahren Baysian Inference und Divisive Normalization und erörtert exemplarisch deren Anwendung in RoboterSystemen. Abschließend werden die Vorund Nachteile einer solchen Anwendung diskutiert.

برای دانلود رایگان متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

Stochastic Bayesian Computation for Autonomous Robot Sensorimotor Systems

This paper presents a stochastic computing implementation of a Bayesian sensorimotor system that performs obstacle avoidance for an autonomous robot. In a previous work we have shown that we are able to automatically design a probabilistic machine which computes inferences on a Bayesian model using stochastic arithmetic. We start from a high level Bayesian model description, then our compiler g...

متن کامل

Bayesian Robots Programming

We propose a new method to program robots based on Bayesian inference and learning. The capacities of this programming method are demonstrated through a succession of more and more complex experiments. Starting from learning of simple reactive behaviors, we present instances of behaviors combination, sensor fusion, hierarchical behavior composition, situation recognition and temporal sequencing...

متن کامل

Design and Construction of a New Capacitive Tactile Sensor for Measuring Normal Tactile Force

This paper presents the design, construction and testing of a new capacitive tactile sensor for measurement of normal tactile force. The operation of proposed sensor has been investigated in ASTABLE and MONOSTABLE circuits. According to the results of these circuits the deviation of ASTABLE circuit results is less than MONOSTABLE circuit results. In addition, the results obtained from ASTABLE c...

متن کامل

Sensor planning for mobile robot localization using Bayesian network inference

We propose a new method of sensor planning for mobile robot localization using Bayesian network inference. Since we can model causal relations between situations of the robot's behavior and sensing events as nodes of a Bayesian network, we can use the inference of the network for dealing with uncertainty in sensor planning and thus derive appropriate sensing actions. In this system we employ a ...

متن کامل

An Introduction to Inference and Learning in Bayesian Networks

Bayesian networks (BNs) are modern tools for modeling phenomena in dynamic and static systems and are used in different subjects such as disease diagnosis, weather forecasting, decision making and clustering. A BN is a graphical-probabilistic model which represents causal relations among random variables and consists of a directed acyclic graph and a set of conditional probabilities. Structure...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2014