Abstract
For robots to navigate effectively in human-shared spaces, they should follow behavioral rules that not only prevent collisions with humans but also reduce uncertainty about the robot’s actions. These rules should be grounded in human collision-avoidance behavior, as humans can often recognize evasive intentions by observing another person’s movements, and robots are expected to exhibit behaviors that allow their evasive intentions to be similarly understood. This paper identifies and characterizes human collision-avoidance maneuvers using a multi-method approach comprising qualitative analysis of video recordings to identify the maneuvers, survey-based analysis to assess self-perceived evasion distances, and computer robot simulator-like human–robot interaction experiments to validate the observed metrics and investigate whether the identified human behavior patterns extend to human-guided robot navigation. A symmetry–asymmetry perspective is then adopted to investigate these maneuvers, and a functional interpretation of the main evasive behaviors is proposed based on their morphological properties and Path Selection Freedom. This interpretation is subsequently translated into human-inspired collision-avoidance strategies for mobile robots through a Behavior Management Module implemented using fuzzy inference mechanisms for maneuver selection and trajectory generation. Beyond the implementation itself, this work establishes a principled basis for transferring the functional principles underlying human collision avoidance to mobile robot navigation, contributing to the development of more natural, understandable, and socially conforming robot behavior in shared human environments.