Motion Planning in Crowds: Proxemics as a Base for a Socially Acceptable Behaviour

When referred to robots, the word intelligence has several facets. If a robot is required to operate in full autonomy, it needs a correct understanding of its environment, of the different spaces that compose it and of their topology; it also needs to detect the different entities that are usually disseminated in the environment, to classify each of them and identify their role in relation to the robot’s mission. An entity of a peculiar kind are human beings. Humans can be simply an hindrance to the robot’s mission, or can be active actors in its execution (e.g., in Cobot application [1] or in shared authority approaches [2] there is a close partnership between humans and robots). Let us now restrict the focus to a specific problem: navigation of an autonomous mobile robot across promiscuous areas, shared with human beings. In cases like this, the actions of a robot are socially acceptable insofar as they reflect the behaviours that a human would generate in similar conditions. The Anthropologist Edward C. Hall first coined the term “proxemics” in oder to describe how humans use the interpersonal space and how they establish non-verbal communication channels tightly related to their interpersonal distance [3]. Hall has identified four zones where thresholds are crossed and behaviours differ: 1. Intimate space (below 450 mm apart), 2. Personal space (450 mm to 1.2 m apart), 3. Social Space (2 m to 3.6 m apart) and 4. Public space (3.6 m to 7.6 m apart). The Intimate space should never be violated except for limited circumstances (e.g., athletes tackling each other, lovers, or parent/children pairs). The Personal space allows for rather intense social interactions (e.g., holding one’s hand, or engaging with friends in face-to-face informal discussions), while the Social space is reserved to more formal interactions (e.g., business transactions). Finally, the Public space is used for interaction with groups (e.g., a speech, a lecture). In this paper, we propose a cognitive architecture to manage social interactions in shared spaces closely inspired to the principle of proxemics.