Publication: Remorobo as the new middleware and framework for heterogeneous multi- robot system_
Date
2023-10-01
Authors
Tareq Gamal Eldin Mohammed Mahmoud
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Middleware for a multi-robot system plays a vital role during the facilitation of effective coordination, cooperation and group task leading towards a common goal and objective. Modern multi-robots are considered a complex distributed system consisting integrated hardware and software components or modules. These modules each comprising sensors, actuators and controllers may need to be interconnected and communicated under different operating system and distinctive communication protocols. The contemporary middleware is designed and developed to solve the interoperability and configuration issues that came along with the module integration. Based on literature survey, multitude of middleware have been introduced by either commercial entities or open-source communities. The wider the adoption of the middleware by the user community (i.e. robotics researchers), the more sustainable will the middleware be receiving support and maintenance. This is crucial for any successful middleware to be accepted as a standard. Often, such case is true only when the middleware is developed by the open-source community. Therefore, factors such as opensource, simulation and hardware support, additional peripheral hardware equipment, compatibility and usability have been studied. The evidential gap is identified leading to the creation of RemoRobo; the new middleware framework suited for multi-robot applications. RemoRobo has been developed with multi-threading and multi-tasking features allowing an almost real-time execution of the peripheral tasks. RemoPillar, forming the backbone of the proposed framework, provides the communication protocol contained herein the command types, message request protocol, task management features imbued with artificial intelligence (AI) features. The developed middleware framework has been field-tested to be applied in a multi-robot-based architecture system which interact with other peripheral robots. These peripheral robots, designed to be customizable, are called Extensions comprising the motion control of a mobile robot (RemoMotion), the synchronizing glowing of LEDs (RemoGlow) and the display of animative facial robot expression for human-robot interaction (RemoFace). This testing phase serves to evaluate the output of the RemoRobo framework, revealing promising results. The average start response time for RemoRobo robots is 29.75 milliseconds, which is significantly faster than the overall start response time of the ROS-based heterogeneous multi-robot system environment which is evidently captured in 50 milliseconds. Among the three test sets, RemoMotion exhibited swift performance, achieving task completion within 22 ms, from command initiation to motion execution and overall task realization. RemoGlow, responsible for LED lighting tasks, accomplished task completion in a mere 34 ms. Moreover, RemoFace effectively rendered a full range of animated facial motions in only 45 ms. These findings highlight the noteworthy speed and efficiency of the new middleware framework.