Page 81 - Artificial Intelligence for the Internet of Everything

P. 81

CHAPTER 4

Active Inference in Multiagent

Systems: Context-Driven

Collaboration and Decentralized

Purpose-Driven Team Adaptation

Georgiy Levchuk, Krishna Pattipati, Daniel Serfaty, Adam Fouse,
Robert McCormack
Aptima Inc., Woburn, MA, United States

4.1 INTRODUCTION

Autonomous intelligent systems are no longer a fancy of science fiction
writers; instead, they are quickly becoming part of our everyday lives. These
devices, from heart monitoring implants to home-heating control systems,
make our lives easier. Commercial technology developers make these
devices “smarter” every day. While most of the currently deployed Internet
of Things (IoT) systems perform simple tasks, like environment monitoring
and human-guided control for smart homes, hospitals, or assembly plants, it
is not difficult to envision a near future in which the intelligence and author-
ity of these devices expand well beyond their current applications.
Most research in the area of IoT intelligence has been focused on the
devices, including hardware–software interoperability (Al-Fuqaha, Guizani,
Mohammadi, Aledhari, & Ayyash, 2015), standards and architectures
(Perera, Zaslavsky, Christen, & Georgakopoulos, 2014), and operational chal-
lenges for individual devices or networks of homogeneous IoT components
(Whitmore, Agarwal, & Da Xu, 2015). However, increasing interdepen-
dencies between component devices, data, physical systems, and human users
prompted researchersandpractitioners to exploretheimplicationsofemergent
device intelligence on broader aspects of our everyday lives (Evans, 2012),
forming the field of the Internet of Everything (IoE). Such studies allow the
development of models to extract the highest potential from multiple autono-
mousandheterogeneousintelligentsystems,includinghuman–machineteam-
ing recently identified as the defense technology of the future (Pellerin, 2015).

76 77 78 79 80 81 82 83 84 85 86