Autonomous Passenger Shuttles (APS) are rapidly becoming an urban public transit alternative. Traversing populous commercial and residential centers, these shuttles are already operating in several cities. In the absence of a human driver and embedded means of communicating the autonomous shuttle's intent, the task of seamlessly navigating crosswalks and pedestrian-friendly zones becomes a challenging pursuit for pedestrians. We contribute to the emerging notion of AV-Pedestrian Interaction by examining the context of autonomous passenger shuttles (APS) in real-world settings, and by comparing four different classes of visual signals -- namely instructional, symbolic, metaphorical, and anthropomorphic -- designed to communicate the shuttle's intentions. Following a participatory methodology involving local residents and public transport service provider, and working within the framework of inflexible road traffic regulations concerning the operation and testing of autonomous vehicles, we conducted a participatory design workshop, a qualitative, and a survey study. The findings revealed differences across these four classes of signals in terms of pedestrians' subjective perceptions. Anthropomorphic signals were identified as the preferred and effective modality in terms of pedestrians' interpretation of the communicated intent and their perceived sense of attention, confidence, and calmness. Additionally, pedestrians' experiences while judging the intention of transitionary vehicular states (starting/slowing) were reported as perplexing and evoked stress. These findings were translated into design and policy implications in collaboration with other stakeholders, and exemplify a viable way for assimilating human factors research in urban mobility.