<p>This work seeks to incorporate existing methodologies for controlling and localizing robotic hardware of different capabilities in order to create an autonomous system of robots to move deliverables within an environment.&nbsp;Our environment is a warehouse; however, we are designing the system with the intention of being able to implement it in any&nbsp;space&nbsp;and to account for any&nbsp;restrictions,&nbsp;whether it be&nbsp;constraints&nbsp;or obstacles in the workspace.&nbsp;The&nbsp;robotic designs that are implemented are modular in design to test different ideas that could be employed within the system and displays the&nbsp;system’s adaptability.&nbsp;Within this&nbsp;work, multiple ground vehicles are working&nbsp;in synchronization&nbsp;with one another&nbsp;as well as&nbsp;a&nbsp;six degree&nbsp;of freedom arm, making&nbsp;the system as dynamic as possible.&nbsp;A high-level task manager assigns&nbsp;priorities&nbsp;to each&nbsp;deliverable, and through decentralized planning, the robots deconflict with one another during execution. Through the adoption of reinforcement learning techniques, our robots create individual path plans that adapt to both static and dynamic obstacles in real time. Contrasting to other works that manage robotic systems in a predetermined&nbsp;or fixed&nbsp;environment, we demonstrate a single, simplified approach that can control and localize modular robots simultaneously in&nbsp;any given&nbsp;setting, including warehouses or factories.&nbsp;A key component to the work is that process&nbsp;planning,&nbsp;and&nbsp;hardware&nbsp;design are&nbsp;occurring&nbsp;in parallel.&nbsp;Traditionally, processes and hardware are&nbsp;designed and tested asynchronously, resulting in a product that fits a narrow&nbsp;objective.&nbsp;By working to design this system in parallel, we are able to create a broadly applicable&nbsp;system of robots to fit the needs of&nbsp;unique end users and circumstances.&nbsp;The localization of the robots is done by using an indoor global positioning system within the set environment. The Indoor GPS hardware is a series of Motion Capture Cameras placed strategically in the environment allowing for a live stream of data being sent to the task manager of all the tracked robots that are marked with Infrared Markers.&nbsp;The live stream data allows for the task manager to operate in real time and send the updated&nbsp;paths&nbsp;to each decentralized system. It does this by&nbsp;utilizing&nbsp;the&nbsp;Robotic Operating System (ROS) to communicate between multiple robots of various capabilities and a single task manager,&nbsp;synchronizing&nbsp;autonomous movement and&nbsp;avoiding&nbsp;other tracked objects.&nbsp;The Robotic Operating System&nbsp;allows us to&nbsp;completely&nbsp;integrate all the micro-systems within the larger system,&nbsp;resulting in&nbsp;the desired&nbsp;autonomous system.&nbsp;&nbsp;</p>

Reinforcement Learning on Modular Robots Using Real-Time Decentralized Deconfliction in a Warehouse Environment

Paper Proceedings: https://asmedigitalcollection.asme.org/IMECE/IMECE2021/volume/85543

ASME 2021 International Mechanical Engineering Congress and Exposition (IMECE2021)

Supporting the need among researchers worldwide for additional online and multimedia content, ASME now offers the ASME Conference Video Collection. This platform will collect and provide long-term access to conference presentation videos, encompassing all technology areas within ASME's technical conference and events program. About 30,000-40,000 engineers across all mechanical engineering and related disciplines participate in ASME conferences each year. This includes authors of approximately 6,000 to 8,000 technical papers published by ASME, the majority of whom also submit video presentations for our virtual conferences. The video collection will be available as an add-on to existing product packages or as a separate subscription.