This paper explores the application of artificial intelligence on edge devices to enhance security in critical infrastructures, with a specific focus on the use case of a battery-powered mobile system for fire detection in tunnels. The study leverages the YOLOv5 convolutional neural network (CNN) for real-time detection, focusing on a comparative analysis across three low-power platforms, NXP i.MX93, Xilinx Kria KV260, and NVIDIA Jetson Orin Nano, evaluating their performance in terms of detection accuracy (mAP), inference time, and energy consumption. The paper also presents a methodology for implementing neural networks on various platforms, aiming to provide a scalable approach to edge artificial intelligence (AI) deployment. The findings offer valuable insights into the trade-offs between computational efficiency and power consumption, guiding the selection of edge computing solutions in security-critical applications.

Advances in Artificial intelligence (AI) and embedded systems have resulted on a recent increase in use of image processing applications for smart cities' safety. This enables a cost-adequate scale of automated video surveillance, increasing the data available and releasing human intervention. At the same time, although deep learning is a very intensive task in terms of computing resources, hardware and software improvements have emerged, allowing embedded systems to implement sophisticated machine learning algorithms at the edge. Additionally, new lightweight open-source middleware for constrained resource devices, such as EdgeX Foundry, have appeared to facilitate the collection and processing of data at sensor level, with communication capabilities to exchange data with a cloud enterprise application. The objective of this work is to show and describe the development of two Edge Smart Camera Systems for safety of Smart cities within S4AllCities H2020 project. Hence, the work presents hardware and software modules developed within the project, including a custom hardware platform specifically developed for the deployment of deep learning models based on the I.MX8 Plus from NXP, which considerably reduces processing and inference times; a custom Video Analytics Edge Computing (VAEC) system deployed on a commercial NVIDIA Jetson TX2 platform, which provides high level results on person detection processes; and an edge computing framework for the management of those two edge devices, namely Distributed Edge Computing framework, DECIoT. To verify the utility and functionality of the systems, extended experiments were performed. The results highlight their potential to provide enhanced situational awareness and demonstrate the suitability for edge machine vision applications for safety in smart cities.

Artificial intelligence (AI) has experienced a recent increase in use across a wide variety of domains, such as image processing for security applications. Deep learning, a subset of AI, is particularly useful for those image processing applications. Deep learning methods can achieve state-of-the-art results on computer vision for image classification, object detection, and face recognition applications. This allows to automate video surveillance reducing human intervention. At the same time, although deep learning is a very intensive task in terms of computing resources, hardware and software improvements have emerged, allowing embedded systems to implement sophisticated machine learning algorithms at the edge. Hardware manufacturers have developed powerful co-processors specifically designed to execute deep learning algorithms. But also, new lightweight open-source middleware for constrained resources devices such as EdgeX foundry have emerged to facilitate the collection and processing of data at sensor level, with communication capabilities to cloud enterprise applications. The aim of this work is to show and describe the development of Smart Camera Systems within S4AllCities H2020 project, following the edge approach.

To meet the demands of a rising population greenhouses must face the challenge of producing more in a more efficient and sustainable way. Innovative mobile robotic solutions with flexible navigation and manipulation strategies can help monitor the field in real-time. Guided by Integrated Pest Management strategies, robots can perform early pest detection and selective treatment tasks autonomously. However, combining the different robotic skills is an error prone work that requires experience in many robotic fields, usually deriving on ad-hoc solutions that are not reusable in other contexts. This work presents Robotframework, a generic ROS-based architecture which can easily integrate different navigation, manipulation, perception, and high-decision modules leading to a faster and simplified development of new robotic applications. The architecture includes generic real-time data collection tools, diagnosis and error handling modules, and user-friendly interfaces. To demonstrate the benefits of combining and easily integrating different robotic skills using the architecture, two flexible manipulation strategies have been developed to enhance the pest detection in its early state and to perform targeted spraying in simulated and field commercial greenhouses. Besides, an additional use-case has been included to demonstrate the applicability of the architecture in other industrial contexts.

espanolLa fabricacion orientada al producto esta completamente alineada con una de las tendencias de la Industria 4.0 que consiste en la integracion de todos los sistemas de produccion. Esto implica una transicion desde un punto de vista tradicional de los procesos, centrados en lineas de produccion con el objetivo de reducir costes, a uno mas flexible capaz de producir productos personalizados. Por otra parte, los Sistema Multi Agente se caracterizan por su caracter inteligente y distribuido, asi como por su capacidad de negociacion para conseguir sus objetivos y se han aplicado ampliamente en entornos de fabricacion. Los trabajos relacionados demuestran su aplicabilidad en entornos flexibles. Por ello, son candidatos muy adecuados para implementar planes de fabricacion cambiantes y personalizados. Sin embargo, un aspecto clave es ofrecer metodologias y herramientas que soporten la implementacion de dichos sistemas. En este articulo se presenta una arquitectura multi-agente para fabricacion inteligente y se ilustra su flexibilidad para el caso de fallos de recursos de fabricacion, dando pautas y plantillas para la implementacion de aplicaciones concretas EnglishProduct oriented manufacturing is aligned with one of the Industry 4.0 trends consisting of integrating all production systems. This implies shifting from a traditional view of manufacturing processes, focused on production line in order to reduce costs, to a more flexible and customized product manufacturing. On the other hand, Multi Agent Systems (MAS) are characterized by their smart and distribute nature as well as their negotiation capacity in order to achieve their objectives. MAS systems have been widely used in flexible environments where related works validate their use. That is why they are a suitable approach for implementing customized and dynamically changing manufacturing plans. However, a key aspect is to offer methodologies and tools for supporting the implementation of such systems. In this paper, a flexible manufacturing supporting architecture is presented and its flexibility is illustrated for the case of resource failures. Procedures and templates are also given in order to implement specific applications

European agriculture is facing numerous challenges such as population growth, climate change, resource shortages and increased competition, and hence the challenge today is to produce “more with less”. Greenhouses protect crops from adverse weather conditions allowing year-round production and integrated crop management approaches provide better control over pests and diseases. However, without adequate controls the intensification of greenhouse crop production can create favourable conditions for devastating infestation. The Horizon 2020 (H2020) GreenPatrol project has developed an innovative and efficient robotic solution for Integrated Pest Management in crops, which has the ability to navigate inside greenhouses while performing early pest detection and control tasks in an autonomous way. The main developments include precise positioning to allow autonomous navigation and provide accurate and detailed pest maps in greenhouses (light indoor environments), perception with visual sensing for on-line pest detection, and strategies for manipulation and motion planning based on pest monitoring feedback. This paper presents an overview of the GreenPatrol system, including the localization and navigation functions and the pest detection abilities, and shows results of real-time demonstrations of the system in a representative environment.

This work presents the team AutonOHM which won the RoboCup@Work competition in Montreal 2018. The tests and main changes of the 2018 world cup competition are presented and a detailed description of the team’s hardware and software concepts are exposed. Furthermore, improvements for future participations are discussed.

The design and operation of manufacturing systems is evolving to adapt to different challenges. One of the most important is the reconfiguration of the manufacturing process in response to context changes (e.g., faulty equipment or urgent orders, among others). In this sense, the Autonomous Transport Vehicle (ATV) plays a key role in building more flexible and decentralized manufacturing systems. Nowadays, robotic frameworks (RFs) are used for developing robotic systems such as ATVs, but they focus on the control of the robotic system itself. However, social abilities are required for performing intelligent interaction (peer-to-peer negotiation and decision-making) among the different and heterogeneous Cyber Physical Production Systems (such as machines, transport systems and other equipment present in the factory) to achieve manufacturing reconfiguration. This work contributes a generic multi-layer architecture that integrates a RF with a Multi-Agent System (MAS) to provide social abilities to ATVs. This architecture has been implemented on ROS and JADE, the most widespread RF and MAS framework, respectively. We believe this to be the first work that addresses the intelligent interaction of transportation systems for flexible manufacturing environments in a holistic form.

Product oriented manufacturing is aligned with one of the Industry 4.0 trends consisting of integrating all production systems. This implies shifting from a traditional view of manufacturing processes focused on production line in order to reduce costs, to a more flexible and customized product manufacturing. On the other hand, Multi Agent Systems (MAS) have been proved to be a suitable way to fulfill these requirements. However, a key aspect of the use of novel technologies is to offer methodologies and tools for supporting the implementation of such systems. In this sense, this paper uses Model Driven Engineering and MAS technology to propose an architecture that is able to launch and execute a manufacturing execution plan. It focuses on the information models managed by architecture agents that can be customized to particular manufacturing plants as well as on the definition of agent templates.

This paper presents the team AutonOHM which won the RoboCup@Work competition in 2017. The tests to be performed during the RoboCup@Work 2017 competition are presented and a detailed description of the team’s hardware and software concepts are exposed. Furthermore, improvements for future participations are discussed.

This paper presents the team AutonOHM which won the RoboCup@Work competition in 2017. The tests to be performed during the RoboCup@Work 2017 competition are presented and a detailed description of the team's hardware and software concepts are exposed. Furthermore, improvements for future participations are discussed.

The aim of this paper is to present a decentralized control architecture for an autonomous transportation system in the manufacturing facility of the future. Each component in the factory (machine, robot, operator ... ) is represented as an individual agent on the cloud and makes autonomous decisions based on the information exchanged with other agents. Production machines control their own material replenishment by contracting the services of Autonomous Transportation Vehicles (ATV) over the cloud. Moreover, two control panels for human synergy on strategical and operational layers for monitoring and interacting with the system have been provided. Based on new trends on the industry 4.0 revolution, this paper develops an intelligent communication architecture between machines, robots and humans. This novel architecture makes the system robust, flexible and scalable. The requirements for such an architecture have been first defined and then validated via a series of theoretical cases and two experiments where communication and hardware failures have been triggered.

The aim of this paper is to present a low-cost Autonomous Transport Vehicle (A TV) for the transportation of material in an existing manufacturing facility. To avoid high costs of layouts and structural modifications, the ATV has to function with minimal changes to the factory. An important part of the transportation process is the docking maneuver: the action of picking up material containers from the storage areas known as supermarkets. This paper proposes three different solutions based on low-cost sensors for this docking maneuver task. First, a line-following method is introduced. To avoid the utilization of markers, the remaining solutions depend exclusively on objects available in the supermarkets. For the second approach, the common blue boxes for carrying materials are used. Third, the rails where material boxes are collected on rollers are utilized for a novel docking guidance approach. Accuracy tests of the three methods proposed expose the possibilities and limitations concerning the range of application. The box detection approach is unsuitable for current supermarkets because of its inaccuracy over larger distances. The line-following approach provides good accuracy, but requires changes to the existing supermarkets. Finally, if global localization provides the correct initial pose, the rail detection has demonstrated an accuracy similar to the line detection approach without the need of additional markers.

This publication describes a 2D Simultaneous Localization and Mapping approach applicable to multiple mobile robots. The presented strategy uses data of 2D LIDAR sensors to build a dynamic representation based on Signed Distance Functions. Novelties of the approach are a joint map built in parallel instead of occasional merging of smaller maps and the limited drift localization which requires no loop closure detection. A multi-threaded software architecture performs registration and data integration in parallel allowing for drift-reduced pose estimation of multiple robots. Experiments are provided demonstrating the application with single and multiple robot mapping using simulated data, public accessible recorded data, two actual robots operating in a comparably large area as well as a deployment of these units at the Robocup rescue league.

This team description paper describes the participation of Team AutonOHM in the RobuCup@work league. It gives a detailed description over the team and the previous achievements. Furthermore, improvements to hardand software for the participation in future competitions are discussed.