Reliable Perception for Heterogeneous Mobile Robot Fleets in Intra-factory Logistic Tasks.

Dimitrios Arapis

Research output: Book/ReportPh.D. thesis

32 Downloads (Pure)

Abstract

The latest industrial transformation, often referred to as Industry 4.0, has been a focus of the corporate sector and the academic world for more than a decade. Breakthroughs in digitalization, artificial intelligence, autonomous robots, and cloud infrastructure have paved the way for a new era of smart manufacturing. Pharma 4.0 is the integration of these technologies within the pharmaceutical industry. Currently, three out of four companies in the pharmaceutical manufacturing domain use robots, with a projection of an increase in adoption. The increased adoption of mobile robots is a result of their declining price and their increasing range of functionalities. Mobile robots are undergoing a transformative evolution that improves their autonomy, adaptability, and collaboration. To efficiently accomplish a diverse range of intralogistic tasks, the ability to use the unique characteristics of different robots is crucial. However, this heterogeneity also adds complexity to the system design, necessitating the development of tailored solutions to effectively integrate and manage different robotic capabilities. These tailored solutions require customized algorithm designs,
collection of process-specific data, and perception mechanisms that will allow robots to infer the world around them. The focus of this dissertation is to expand the perception capabilities of mobile robots in intralogistics. To achieve this goal, advances have been achieved through four key developments: (i) designing a method for depth estimation and completion tailored for devices with limited computational resources, (ii) investigating multi-task learning and task-balancing techniques for dense prediction tasks in mobile robotics, (iii) producing a synthetic dataset for mobile robotics applications in warehouses, and (iv) designing a perception module centered around human presence in industrial environments. Advances in mobile robot perception have the potential to enhance robots’ operation in intra-logistics tasks, thereby strengthening their role and adaptation in future manufacturing.
Original languageEnglish
PublisherTechnical University of Denmark
Number of pages138
Publication statusPublished - 2024

Fingerprint

Dive into the research topics of 'Reliable Perception for Heterogeneous Mobile Robot Fleets in Intra-factory Logistic Tasks.'. Together they form a unique fingerprint.

Cite this