Motion Categorisation: Representing Velocity Qualitatively

Juan Purcalla Arrufi , Alexandra Kirsch (1, 2, 3)

1 Institute for Advanced Studies
2 Intelligent Autonomous Systems
3 Human-Computer Interaction and Artificial Intelligence

Categorising is arguably one of the first steps in cognition, because it enables high-level cognitive processing. For a similar reason, categorising is a first step—a preprocessing step—in artificial intelligence, specifically in decision-making, reasoning, and natural language processing. In this paper we categorise the motion of entities. Such categorisations, also known as qualitative representations, represent the preprocessing step for navigation problems with dynamical obstacles. As a central result, we present a general method to generate categorisations of motion based on categorisations of space. We assess its general validity by generating two categorisations of motion from two different spatial categorisations. We show examples of how the categorisations of motion describe and control trajectories. And we establish its soundness in cognitive and mathematical principles.

Domaines: Intelligence artificielle [cs.AI]

Juan Purcalla Arrufi, Alexandra Kirsch. Motion Categorisation: Representing Velocity Qualitatively. Cognitive Systems Research, 2018, 52, pp.117 - 131. ⟨10.1016/j.cogsys.2018.06.005⟩. ⟨hal-01845141⟩