Designing Socially Aware Robot Motion

Autonomous mobile robots (AMRs) have increasingly been used in dynamic, human-populated environments. The robots' ability to navigate not just safely, but also socially in the sense of respecting social rules, becomes therefore essential. This dissertation addresses the emerging field of Socially Aware Navigation, presenting novel motion planning frameworks that extend Dynamical System Modulation (DSM) with proactive capabilities to bridge the gap between reactive safety and social compliance. Two methodological approaches are presented. The first one introduces Social Inference Model. Inspired by the Social Force Model, it interpretes the environmental social forces on the robot as additional velocity dynamics. The second method leverages nonlinear opinion dynamics to model the robot's passing-side decision as a bifurcating dynamic process. Furthermore, proxemic zones and specific speed profiles are incorporated into the DSM design to explicitly investigate the impact of robot behavior on perceived human comfort. Validated through extensive simulations and a real-world user study, this work offers insights into human perception of robot behaviors. Subjective and objective findings demonstrate that considering asymmetric personal proxemic spaces and maintaining consistent speed profiles enhance perceived human comfort, while the benefits of proactive behavior are shown to be context-dependent.