This workshop aims to develop a roadmap for understanding the impact of robotics on human employment within the context of Industry 5.0. The workshop explores key factors in this transition, including the risks and benefits of digital automation, ethical considerations regarding worker skills and well-being, and protective measures for workers and consumers. Focusing on the theme of sociocultural risks: devaluation of human labor, the workshop aims to reflect on traditional notions of work, productivity, and the value of human labor in the era of advancing robotics and automation. It challenges the concept of labor devaluation, assesses its impact on worker recognition and efficiency, and questions meritocratic paradigms. To facilitate interdisciplinary dialogue, the workshop engages philosophers and engineers in addressing ethical dilemmas in technology development. It explores the social significance of human effort and movement when interacting with autonomous robots and occupational exoskeletons. By examining these technologies' adaptability and role in human-machine interaction, especially in workplaces, the workshop intends to contribute to discussions on technological advances and their societal implications.

The presentation aims to explore the relationship between work dynamics, technological innovation, and social cohesion, starting from ethical considerations related to the integration and/or replacement of human labor by digital automation systems, included robots. It discusses the historical evolution from Industry 1.0 to 5.0, outlining each phase's characteristics and impacts. It presents the interconnection between an analysis of meaningful work and the prospect of transitioning to Industry 5.0, highlighting its focus on human-centered production and evaluating the significance of labor in a scenario of progressively evolving conventional employment. Central to this exploration is the concept of 'meaningful work'—work that transcends mere production, enriching personal and social experiences and contributing to individual fulfillment and societal fabric. Finally, it delves into the debate surrounding the concept of a 'new Digital Social Contract,' as highlighted by the High-Level Expert Group on the Impact of Digital Transformation on EU Labour Markets report (2019), reflecting on the challenges that this evolving digital landscape poses to our foundational societal agreement, reevaluating the role and temporal contribution of individuals to labor and examining the impact for the social fabric of the labor market.

This presentation examines the effects of exoskeletons in the workplace on workers’ affectivity, introducing the novel notion of 'body invasion' as an extension of 'mind invasion'. Mind invasion critiques the user-resource model in extended mind debates and it refers to the ways in which it is not exactly my individual decision to employ a mind tool to pursue the goals I chose for myself, but rather “forms of pervasive framing and molding effected by aspects of technical infrastructure and institutional realities”. Similarly, we propose 'body invasion', which refers to the potential impact of wearable technologies like exoskeletons. By changing the embodiment of workers in ways that might go beyond the fulfillment of goals the worker chooses for herself, these technologies can reshape affective repertoires, especially in the interactions with other workers. For instance, while they can enhance performance, they may also raise expectations, increasing anxiety. This change can lead to new bodily affective styles and unexpected affective atmospheres in the workplace. The presentation shows that while these technologies offer benefits, their potential risks, especially in altering affectivity and embodiment, must be considered when designing and regulating these technologies.

One major barrier to adoption for occupational exoskeletons is the limited evidence that the technology is effective, acceptable, and usable by workers in the target scenarios. In addition to the specific characteristics of any device, the selection and proper definition of the application could strongly influence the successful deployment of the technology in industrial scenarios. In this talk, we will discuss the main biomechanical and safety requirements that have been considered for the design of two occupational exoskeletons (OEs) by IUVO and COMAU (one upper-limb exos and one lumbar exos), as well as their validation path. Since the beginning of the development phase, OEs has gone through a continuous testing process with human subjects to assess their effectiveness and potential undesired effects on the users. In addition to initial in-lab tests, we carried out “on-site” tests with experienced workers, with the final goal to verify the effects of using the device in real or realistic application scenarios and thus, identify suitable applications for the successful deployment of the technology. On-site evaluation studies included a comprehensive set of metrics combining instrumental parameters (electromyographic recordings on the main upper-limb muscles) and perception-related parameters (usability and acceptability questionnaires). 

To be deployed in unstructured environments, autonomous systems such as robots need the ability to learn motor skills autonomously, through continuous interaction with the environment, humans and other robots. Although classically built on rigorous control theory, mathematical and theoretical computer science methodologies, more recently data-driven learning methods, such as Deep Learning (DL) and Reinforcement Learning (RL) have been demonstrated as powerful technologies for developing AS. Still, most of the practical applications exist in carefully structured settings where i) there exists enough data to train the models, ii) one can structure the search problem efficiently, and iii) there is a computational infrastructure and/or many robots to run large scale experiments. The last point in particular is valid for only a handful of labs in the world and even in their case, mostly in collaboration with a handful of companies.  --   Despite the progress, four open problems are: i) multi-task and transfer learning, ii) learning with little data, iii) learning when a reward function or the desired outcome of a task is difficult to define, and iv) learning in a continual/non-episodic manner. All these four are extremely important in robotics given that a robot acts in and interacts with the physical world that is dynamic and unstructured. A robot perceives the world with a multitude of sensors, is expected to build and update a model of the world incrementally and continuously over time and use it to make decisions, plan actions, fulfill useful tasks and generalize between tasks that are similar in nature. For robots to become safe and robust, it is essential to understand how learning algorithms for discrete sequential decision-making can interact with continuous physics based dynamics. The research problems considered in this project span over several artificial intelligence research areas: computer vision, machine learning, and robotics. Despite the important progress in all three areas, there are still no robots that exhibit anything close to human ability to acquire and learn complex behaviors. In our recent work, we build upon two major recent developments in the field, Diffusion Policies for visuomotor manipulation and large pre-trained multimodal foundational models to obtain a robotic skill learning system. The system can obtain new skills via the behavioral cloning approach of visuomotor diffusion policies given teleoperated demonstrations. Foundational models are being used to perform skill selection given the user's prompt in natural language. Before executing a skill, the foundational model performs a precondition check given an observation of the workspace.