Summary:
Robotics and Internet of Things (IoT) have been experiencing rapid growth nowadays. IoT nodes are significantly enhanced with many different features. One of the most important is the mobility capabilities, given by the noticeably huge growth of UxV (UxVs- x stands for a different type of environment, i.e. ‘s’ stands for sea, ‘a’ for air and ‘g’ for ground) area. The idea is the assumption of a drone as a mobile sensor, that can be deployed wherever the experimenter wants. Some more characteristics that make the unmanned vehicles a very tempting decision as IoT nodes are the decision-making ability without human interaction, endurance, re-programmability and capability of multimedia streaming. These characteristics make drones an option for use cases of surveillance, security monitoring, and supporting crisis management activities. For instance, a UGV equipped with a high-definition camera and running an algorithm of object recognition can serve the purpose of border surveillance.
In this thesis, a framework that implements a network quality based decision-making process is developed. This framework adapts the information flow between the UxV and the Ground Control Station (GCS) based on network quality metrics (such as packet error rate etc.) and the principals Optimal Stopping Theory (OST). The goal of this framework is to ensure the optimal delivery of critical information from UxV to GCS and vice-versa. If the network behaves optimally then there is no limitation on the information flow, but if the network is saturated or overloaded restriction rules are applied. The proposed model introduces two optimal stopping time mechanisms based on change detection theory and a discounted reward process.
To support the implemented framework, an experimental environment has been set up and also a series of experiments with very promising results. As a mobile IoT node, a TurtleBot has been used, along with an XBOX Kinect sensor (RGB camera and depth sensor) and a Raspberry Pi running Robotic Operating System (ROS) and Apache Kafka pub-sub system with ultimate purpose the communication between the TurtleBot and the GCS.
Keywords:
Robotics, Network reliability, Decision making, IoT, Optimal Stopping Theory, Unmanned Vehicles
