A Decentralized Coordination Algorithm for Patrolling and Target Tracking in Internet of Robotic Things (IoRT) using Dynamic Waypoints and Self-triggered Communication

Abstract

The Internet of Robotic Things (IoRT) deals with autonomous objects such as robots and sensors together. It can be used for efficient patrolling and target tracking; however, mobile sensors (robots) must coordinate to decide their optimal actions. This paper proposes a decentralized coordination strategy that can enable multiple robots to perform both patrolling and target tracking in the deployed scenario. The basic idea is that the whole site is divided into local zones, and a single robot is deployed in each zone. All robots are logically divided into two groups, namely category C1 and category C2 robots. A dynamic waypoint generation algorithm is proposed to assist category C1 robots in the perimeter patrol. It produces the waypoints such that certain locations can be prioritized and intruders cannot predict the patrolling trajectory. Category C2 robots are responsible for area patrolling within the zone. Here, we also propose a strategy such that the places with a high probability of unusual acts can be visited frequently. We use the distributed Extended Kalman filter (EKF) to estimate and predict the position of the targets. Each robot has a self-triggered communication mechanism to share the necessary information with the neighbors, such as the estimated position of the intruder, EKF parameters, asking for help, etc. We have also developed an Internet of Thing (IoT) based web application to monitor and control the robots. In this app, robots subscribe to the server for necessary information and commands and publish their position, patrolled area, intruders’ position, battery status, etc., to the server for real-time monitoring and control. The proposed solution is validated through simulations in the Robot Operating System (ROS) and Gazebo. The results show the patrolling and target tracking performance using idleness and error in the target’s estimated position, respectively, as metric.