CORESENSE is a project aimed at solving the problems faced by autonomous robots in open environments. The current trend towards increased autonomy in robots has led to augmented problems of reliability, resilience, and trust. This is due to the robots’ lack of understanding of their surroundings and their role in it. CORESENSE aims to address this by developing a cognitive architecture for autonomous robots that is based on a formal concept of understanding, supporting value-oriented situation understanding, and self-awareness.
Our project has four objectives, including the formulation of a theory of awareness based on a systemic concept of understanding, implementation of a reference architecture and engineering toolbox for aware autonomous robots, and the use of the theory and technology in the construction of autonomous robots with increased capabilities.
CORESENSE differs from other AI-driven approaches in that it is not data-driven or human-inspired. Instead, we use a formalistic, model-based approach that is based on principles, making it applicable to a broad range of technical systems. Our aim is to provide a solution that will lead to improved performance, flexibility, reliability, and user-experience in autonomous robots.
Overview of the CORESENSE Work Packages
CORESENSE activities will be implemented under 9 different Work Packages.
Work Package 1: Understanding and Awareness Theory
Work Package 1 of the CORESENSE project focuses on developing the theoretical foundation for the project. It aims to formalize the theory of understanding and awareness, and to consolidate the concepts that underlie these theories in the form of formal ontologies. These ontologies will be used to represent world/agent states at runtime and to drive self-awareness and self-control. They will also be used in the engineering process to derive model-based engineering metamodels for system construction. WP1 will develop three formal ontologies: Intentional Systems Ontology (ISO), Understanding and Awareness Ontology (UAO), and CoreSense Ontology (CSO). These ontologies will capture different aspects of the CoreSense architecture and enable the system to capture the essential concepts necessary for self-awareness and self-control.
Work Package 2: Awareness Architecture
Work Package 2 of the CORESENSE project focuses on developing a reference architecture for robot awareness -CoreSense-, which includes three parts: runtime system, understanding core, and awareness system. The runtime system is a software infrastructure that deploys, connects, monitors, and controls modules using the ROS2 platform and tooling from WP5 and previous EU-funded projects. The understanding core computes meanings on demand and the awareness system generates awareness by using sensing modules, the core, and the mental model. The architecture will be developed through conceptual analysis, specification, software development, and refinement.
Work Package 3: Cognitive Modules and Structures
Work Package 3 of the CORESENSE Project aims to develop generic cognitive modules and structures to be used in the testbeds and beyond. The package involves the analysis of needs for cognitive capabilities from various sources, the selection of capabilities, and the specification of necessary modules and structures. The implementation of the cognitive modules will be based on well-known algorithms and available software, and some specific developments may be required. The structures include Anticipation, Navigation, Reconfiguration, Explanation, Control, and Diagnosis. The project will use system models to follow the life-cycle doctrine and reuse existing code as much as possible to reduce effort.
Work Package 4: System Lifecycle and Toolchain
Work Package 4 of the CORESENSE Project aims to develop a robotic control systems life cycle and a model-driven engineering toolchain to support the life-cycle activities. The robotic system engineering lifecycle will be based on the ISO/IEC/IEEE 24641.2 standard and tailored to the CoreSense architecture. The toolchain will support the design and management of the robot system model and its components, including the cognitive modules, and will facilitate the packaging and deployment of the solution. The development of the toolchain will involve the specification of a SysML v2 profile, a system model graphical editor, a cognitive module designer and configurator, a ROS2 model transformation tool, a CoreSense modelet designer and generator, and a runtime system monitor and debugger.
Work Package 5: ROSification of Software Assets
Work Package 5 in the CORESENSE project is focused on the “ROSification of Software Assets.” The objectives are to support the use of ROS across the project, instrument the ROS2 platform for sensing and actuation, and package the project results in a format suitable for the ROS ecosystem. This work package will be divided into three tasks: T5.1 ROS Development Support, T5.2 ROS Platform Instrumentation, and T5.3 ROSification of CoreSense Results. The first task will consolidate code development guidelines and provide support to other work packages. The second task will instrument the ROS2 platform with necessary software sensors and actuators for self-awareness and control. The final task will prepare CoreSense results for the ROS community, including proper documentation, shared repositories, and tutorials.
Work Package 6: Testbed: Manufacturing robots
Work Package 6 of the CORESENSE project aims to demonstrate the augmented flexibility and autonomy of robots in manufacturing cells through the construction and testing of two industrial robotic demonstrators. The first demonstrator is an on-demand mobile manipulator, which is capable of executing discrete assembly tasks and responding to machine maintenance calls. The second demonstrator is an industrial robot cell for manufacturing large load-bearing composite structures, equipped with multiple exchangeable end effectors. The project will integrate perception and cognitive modules from WP3 into the testbeds, enabling flexible robot behavior and communication with operators through augmented reality. The testbed will be evaluated in two manufacturing scenarios, and if successful, elements of the testbed could potentially be developed into a commercial solution.
Work Package 7: Testbed: Inspection drones
Work Package 7 of the CORESENSE Project is focused on the demonstration of the improved resilience of a group of inspection drones by means of the awareness architecture. A testbed will be set up in the form of an outdoor inspection mission performed by a swarm of drones in a distributed manner. The inspection of a Photovoltaic Power Plant will serve as a testbed scenario. The goal of the testbed is to demonstrate the resilience of the system and to achieve a complete inspection task even when there are failures or external disturbances. Tasks within WP7 will involve conducting a requirement analysis, defining the inspection mission, specifying the collective awareness architecture, setting up the drone swarm and control code, and evaluating the testbed through several failure scenarios.
Work Package 8: Testbed: Social robots
Work Package 8 of the CORESENSE Project focuses on testing social robots in a domestic environment. The testbed is based on the Robocup@home competition and will demonstrate the enhanced performance of social robots in common domestic tasks, as well as the increased explainability of AI systems. The testbed will evaluate CoreSense in two dimensions related to human values and alignment and two dimensions related to situational awareness and performance. The social testbed will be developed by PAL, URJC and UPM, and will be evaluated by all partners, with feedback delivered to the theory and technology threads of the project.