Physical Human-Robot Interaction in Anthropic Domains: Safety and Dependability

In this paper we describe the motivations and the aim of the EURON-2 research project “Physical Human-Robot Interaction in Anthropic Domains” (PHRIDOM). This project, which moves along the lines indicated by the 1 IARP/IEEERAS Workshop on Technical Challenge for Dependable Robots in Human Environments [1], is about “charting” the new “territory” of physical Human-Robot Interaction (pHRI). To ensure these goals, the integration competences in control, robotics, design and realization of mechanical systems, human-machine interaction, and in safety-dependability of mechatronic systems is required. The PHRIDOM Consortium is composed of 5 partners from 3 different European countries. I. GENERAL DESCRIPTION AND SCOPE OF PHRIDOM The PHRIDOM project is about exploring the relatively new research field of physical Human-Robot Interaction (pHRI). In writing the project, we have often used metaphorically the language of pioneering explorations of the nineteenth century. Thus, we plan in this project to explore the uncharted “territory” of pHRI and contribute to prepare an “atlas” for it (see e.g. fig. 1). Its ”geographical features” mainly consist of • Applications(Destinations): tens of examples of intelligent machines embedded in anthropic domains i.e. environments shared by machines and humans, working together elbow-to-elbow, or even more closely; • Requirements(Viability Conditions): safety, dependability, reliability, failure recovery, performance; • Technologies(Via Points): sensors, actuators, mechanics, control, SW architectures; • Systems(Pathways): connecting crucial components and leading to technological solutions to applications, while fulfilling the requirements; • Competences(Crews): the centres of excellence among academic and industrial groups from which a successful research crew has been recruited. The planned “atlas” will be useful in the near future to navigate in this new research field using knowledge Fig. 1. The aim of the PHRIDOM project is to explore the new territory of physical Human-Robot Interaction. accumulated so far by pioneers to provide directions to new explorers who want to reach out for new, unexplored applications. It is in this sense a preparatory project, following which we would expect a large number of other projects to be able to make headway into the bushes of a difficult, high-responsibility, yet fascinating research domain. A. Applications involving pHRI Central to PHRIDOM is the highly challenging domain of human-centred robotics, where machines have to closely interact with humans. Applications of robotics in domains such as medical, domestic, public-oriented service, personal assistance and home care are often cited as examples of interesting avenues for development. Applications of intelligent machines that work in contact with humans are however more, and more general, involving e.g. haptic interfaces and teleoperators, cooperative material-handling, power extenders and such high-volume markets as rehabilitation, physical training, entertainment (see e.g. [2], [3], [4], [5]). All the above applications involve human-robot interactions where the person may be a non-professional user or a bystander. Unlike the industrial robotics domain where the workspace of machines and humans can be segmented, machines of these types must, by definition, have physical contact and interaction with the user. The PHRIDOM project will enquire in the above and other fields, scouting for new applications who promise the best return-on-investment, be such return in terms of societal wellness (as e.g. for assistive or public service robotics) or industrial markets (as e.g. in entertainment or training machines) reporting on the application-specific needs in terms of technology and requirements. B. Requirements in pHRI Robots designed to share an environment with humans must fulfil different requirements from those typically met in industry. It is often the case, for instance, that accuracy requirements are less demanding. On the other hand, a concern of paramount importance is safety of the robot system. “A robot may not injure a human being...”, or, to rephrase the famous quote, under no circumstances should a robot cause harm to people in its surroundings, directly nor indirectly, in regular operation nor in failures. Failures in the mechanics or control should happen rarely, if ever. This entails degrees and standards of reliability that must be rethought. From a system viewpoint, however, a pHRI machine must be considered as part of unpredictably changing anthropic environments. From this point of view, “failures” are events (e.g., contacts with a person, unexpected changes of an user’s mind, even users’ mistakes) that cannot be ruled out in principle, and must rather be faced by suitable policies. The need hence arises for fault detection, and for graceful fault management and recovery. In general, pHRI applications also raise critical questions of communication and operational robustness. All these aspects can be captured by the concept of dependability [1], a crucial focus