At the online Digital Meeting for Conversation Analysis, Oct. 31st-Nov. 4th 2022, Pelikan presented on “How an autonomous shuttle bus could sound in interaction”.
Sound, often in the form of honks, is an important interaction device for managing accountability on the road (Deppermann et al., 2018; Laurier et al., 2020). Autonomous shuttle buses are heavily restricted in their ability to coordinate with other traffic participants: despite looking like small buses they drive on pre-programmed routes like a tram on invisible tracks. With their ambiguous affordances for interaction, they are difficult to interact with, which impairs smooth traffic flow and may lead to accidents (Pelikan, 2021). Following a research-through-design approach (Zimmerman & Forlizzi, 2014) the goal of this work is to explore how the bus could communicate its limited interaction capabilities through different types of sound, including continuous jingles, motor hums (Moore et al., 2017), human-inspired vocalizations (Keevallik & Ogden, 2020) and auditory icons (Gaver, 1986) like bell and honk sounds. The paper builds on a corpus of video recordings of Wizard-of-Oz controlled (Porcheron et al., 2020) iterative sound tests in live traffic on a university campus, during which a researcher (the wizard) sat inside the bus and triggered different sounds through a speaker on the outside of the bus whenever it got close to cyclists and pedestrians. The video corpus features four different camera perspectives showing people outside on the road and how they orient towards the shuttle as well as people inside the bus including the researcher and a legally required safety driver, who is also a professional bus driver. Taking an ethnomethodology and conversation analytic (EMCA) approach that focuses on the sequential unfolding of events, each design iteration was informed by analyzing how people oriented to the sounds on a moment-by-moment basis. First, motor hums and human-inspired vocalizations lead to minimal reactions at best and are commented on as too unclear in tsheir origin and communicative purpose by the safety driver. Second, bell-inspired “ding” sounds are relatively short and can be repeated in a rhythmic manner. Traffic participants tend to treat them as warning sounds directed at specific parties. Third, jingles persist continuously and remind of the sound of local ice cream trucks. They seem to be treated as accounts for why the bus is there and invite (playful) interaction with it. The paper discusses the interactional function of different sounds and explores how EMCA can contribute to the design of robot sound that facilitates interaction with humans.