European Commission logo
español español
CORDIS - Resultados de investigaciones de la UE

Collaborative Intelligence for Safety Critical systems

Periodic Reporting for period 1 - CISC (Collaborative Intelligence for Safety Critical systems)

Período documentado: 2021-01-01 hasta 2022-12-31

Collaborative Intelligence for Safety Critical systems is core to the European declared 'human-centric' approach to AI that requires a human capital able to prepare for the socio-economic changes brought about by AI. While the ease of collecting and using field data with AI is increasing, few understand the importance of fully considering how to interface AIs with the humans that are supposed to use them in order to realise the anticipated benefits, and even fewer know how to address these new types of human-machine collaboration and their legal and ethical aspects, In Collaborative Intelligent systems, for instance, humans need to perform three crucial roles. They must train machines to perform certain tasks; explain the outcomes of those tasks, especially when the results are counterintuitive or controversial; and they must sustain the responsible use of machines (by, for example, preventing robots from harming humans). On the other side AI can amplify our cognitive strengths such as filter data to provide us with information about the status of a safety critical plant (e.g. distillation column) & suggest possible procedures to cope with plant status upsets. Furthermore AI systems in collaborative robotics (cobotics) can embody human skills to extend our physical capabilities. In these collaborations the end users should not to be subject to a decision based solely on automated processing and there should always be human oversight. The development of Collaborative Intelligence systems requires an interdisciplinary skillset blending expertise in AI with expertise in Human Factors, Neuroergonomics and System Safety Engineering. The CISC training programme will develop Collaborative Intelligence Scientists 1) Using data analytics and AI to create novel human-in-the-loop automation paradigms to support decision making and or anticipate critical scenarios. Designing and implementing processes capable of monitoring interactions between automated systems and the humans destined to use them; (3) Modelling the dynamics of system behaviours for the manufacturing process considering System Safety Engineering; (4) Managing the Legal and Ethical implications of AI algorithms, and the use of physiology recording wearable sensors and human performance data in them.

The main objectives of this ITN are to:
1) train researchers with interdisciplinary skills and intersectoral experience in the field of AI driven automation for Industry 4.0; merging the fields of Human Factors and AI in Automation integrated with system safety engineering and ethical and legal aspects of Collaborative intelligence
2) promote academic-industry collaborations; and
3) foster European excellent science. CISC explicitly aims to address the United Nation Sustainable Development Goals (UN SDGs) 4.B (eliminate gender disparities in education and ensure equal access, including people with disabilities), 5.B
(Promote Empowerment of Women through Technology), 8.2 (Diversify, Innovate and Upgrade for Economic Productivity) and 9.5 (Enhance Research and Upgrade Industrial Technologies).
Maintaining optimal human and system performance is a major concern in Industry 4.0 particularly for safety-critical applications. Failure in the proper integration between automation, intelligent systems and the operators has resulted before in disastrous consequences, where the poor design of the system has led to a reduction of the operator’s vigilance, reduction of situational awareness, information overload and/or loss of ability to manually control the system. It is then paramount for this new stage of collaborative intelligence in industry to prioritize the human-system interaction and communication to increase awareness of each other’s actions and intentions. Indeed, this shift towards a human-centric approach complements existing industry 4.0 methods and contributes to the European Commission’s vision of Industry 5.0 i.e. a manufacturing eco-system that brings benefits to industry, workers and the broader society.

Within the CISC project, different levels of interaction are studied through the use of Live labs. The Live labs are a means to validate research in near-real environments and ensure that the ESRs are exposed to real-world problems. There are multiple Live labs within the project which can be divided into three main classes. First, researchers will focus on direct human robot interaction. In this scenario, the human will oversee or teach a robot to complete a task. CISC will study the programming methods and aim to optimize this teaching process to improve task efficiency and human ergonomics. Secondly, researchers will focus on exploiting the data from a human executed manufacturing process. To do this efficiently, human performance will be modelled considering task complexity. The data generated during the manufacturing process can then be used to improve human performance and signal potential failures and maintenance events. Finally, researchers will study control room tasks, where humans must oversee complex and critical operations. CISC will show how machine learning and data analysis can alleviate stress, predict overloading scenarios and thus aid allow human operators to make judicious decisions under stressful conditions.

In all three scenarios, the human operator remains a central component responsible for the high-level cognitive actions. The CISC project will aim to amplify an operator’s capabilities while maintaining a level of safety and ergonomics. This deliverable will outline the three LIVELABs central to the CISC project. The components, location and activity are described, and research implementation and experiment design are outlined. Each ESR has the opportunity to validate their algorithms/ approach across all LIVELABS however their methods/ approaches are typically tailored to a single main experiment.
The expected results are classified by Live Labs as follows:

Live Lab 1- Human robot collaboration: Live Lab 1 is divided into three different scenarios located in IMR’s pilot factory in Mullingar and in the Faculty of Engineering, University of Kragujevac. This Live Lab will focus on how operators can better interact with robots at various different control levels. The efficiency and safety of these operations will be studied and data concerning human factors, such as attention and comfort etc., collected.

Live Lab 2- Augmenting Human performance: Live Lab 2 is focused on manufacturing operations in a large-scale automotive plant. The project will collect data during manufacturing and model the human operator’s performance versus task complexity. The objective is to exploit this data to enable optimization of human performance while simultaneously predicting anomalies and scheduling maintenance events.

Live Lab 3- Assisting Human decision-making: Live Lab 3 is divided into two different scenarios, firstly utilizing data obtained from a real control room in Yokogawa’s facility servicing the Oil and Gas industry and secondly data collection and exploitation from a controllable simulated environment developed by POLITO (Politecnico di Torino). The objective is to explore alarm flooding events, i.e. when multiple alarms occur simultaneously and how to predict such events and how to ensure the human operator is not overwhelmed.