Academic discourse on Autonomous Vehicle (AV) futures has largely focused on safety, privacy, accountability, travel behavior, land use, and environmental impacts. However, public narratives are predominantly shaped by incumbent regime actors like car manufacturers, often visually depicting AVs as simple substitutions within existing systems rather than radical transformations. To realize the potential of AVs in mitigating negative social, environmental, and economic impacts, more nuanced public discussions and depictions are needed. AV development is deeply embedded within complex sociotechnical systems. Their implementation will be shaped by the imaginaries of various system actors. Alternative imaginaries exist, envisioning AVs not as part of current transportation systems but as components of new, sustainable constellations. Deterministic assumptions about AVs' impact should be avoided, focusing instead on the plurality of existing imaginaries and exploring how we want this technology integrated into our lives. While the concept of AVs is nearly 75 years old, recent advancements in machine learning have transformed their perception. Early visions involved integrated smart cars and highways, but now AVs are seen as independent entities navigating environments through various sensors and computing power. While developers claim superior environmental perception compared to human eyesight, others highlight an ontological gap between the real world and its computer-vision model. From a socio-technical perspective, AVs are often viewed as solutions for issues like traffic fatalities, pollution, congestion, and climate change. Their role extends beyond transport to national economic development, security, and global leadership. However, these promises often ignore transition periods and potential unintended consequences like increased traffic volume, urban sprawl, and inequality. The terminology surrounding AVs is varied, with terms like "driverless cars," "self-driving vehicles," and "autonomous vehicles" prevalent in public forums, while "automated vehicles" is more common in technical documentation. This difference in terminology influences perceptions and empowers certain visions of autonomy while disempowering others. The SAE Levels of Driving Automation provide a framework for categorizing AV technology, but the public narrative often simplifies the progression of automation, neglecting the complexities of human-machine interaction and shared responsibility.

The paper draws upon existing literature on sociotechnical systems, transitions, and imaginaries to build its analytical framework. It references works on the system of automobility, highlighting the car's pervasive influence on societal structures, economic practices, and urban landscapes. The multi-level perspective (MLP) is employed to understand transitions in complex systems, differentiating between sociotechnical landscapes, regimes, and niche innovations. The concept of sociotechnical imaginaries is crucial in understanding how AVs are developed, propagated, and used, illustrating how actors compete for dominant imaginaries to shape the technology's future use. The paper also leverages visual discourse analysis to understand how visual material is used to construct and disseminate sociotechnical imaginaries.

This research uses visual discourse analysis (VDA) to analyze visualisations from two actors: Daimler, a major automotive corporation, and JAJA Architects, a Danish architecture studio. Daimler's data consisted of images from their Global Media website (2015-2019), focusing on depictions of AVs in future urban environments. JAJA Architects' data came from a scenario-planning project, "Copenhagen 2050," exploring AV integration in sustainable transportation. VDA was applied to understand how images construct and disseminate sociotechnical imaginaries. The methodology draws upon semiotics, discourse analysis, and the grammar of visual design. The analysis focused on compositional and social modalities. Semiotic analysis identified first-order signs (denotation) and second-order signifiers (connotation), relating them to themes identified in the niche and landscape levels of the MLP. Compositional interpretation analyzed content, color, spatial organization, and focalisers. The analysis sought to understand how system actors engage with system pressures in their presentations of AVs and their perceived hierarchies of system elements.

The analysis of Daimler's and JAJA Architects' visualisations revealed contrasting approaches to conceiving AV futures. Daimler's visuals, responding to landscape discourses about road safety and urbanisation, prioritized vehicle use and regime stabilization, presenting a future of continued automobile dominance with AVs as a technological fix. Their depictions initially emphasized vehicle autonomy and safety features, later incorporating elements of multimodality and sustainability to mask a fundamentally car-centric vision. The inclusion of elements relating to emerging niche innovations masked their underlying goal of preserving their existing business model and the dominant position of the car. Daimler's visualisations showed a co-evolution with emerging niche innovations and landscape developments, absorbing destabilising forces into their narrative of AVs. Their images adapted to include elements representing multimodality, sustainability, and emerging niche innovations to create a seemingly forward-thinking approach. However, their actual goal was to maintain their business model by adapting their messaging rather than fundamentally altering their business model. In contrast, JAJA Architects' visualisations, shaped by liveability discourses, presented a counter-sociotechnical imaginary that prioritized urban life and a multimodal transportation system. Their depictions showed AVs as a supporting component, not the dominant mode, emphasizing public transport and the re-appropriation of road space for community use. They focused on a transformation of urban space, promoting cycling, pedestrian areas, and a strong sense of community. Their depictions highlighted sustainability and liveability as critical components of the future, integrating AVs into a system that actively reduced the dominance of the automobile. The analysis highlighted the non-neutral nature of visualisations and the lack of technological determinism in AV development, influenced by actors' values and business models. Daimler’s approach demonstrates an attempt to maintain the dominance of the car within the system of automobility, while JAJA’s approach suggests a reconfiguration of the system that prioritizes urban liveability and multimodal transport. The findings show that visualisations are tools used by different system actors to promote different sociotechnical imaginaries, affecting transition pathways and policy considerations.

The findings demonstrate the power of visual representations in shaping perceptions of AV futures and influencing policy decisions. The contrasting approaches of Daimler and JAJA Architects exemplify how different system actors frame AVs based on their positions and business models. Daimler's 'business as usual' approach, while adapting to landscape pressures, ultimately aims to maintain the car's central role. JAJA Architects, on the other hand, envisions a future where AVs are integrated into a broader, more sustainable and liveable urban system. The research highlights the political nature of AV development, showcasing how visualisations are deployed to shape perceptions and advance specific agendas. The depoliticization of discussions surrounding AVs is also noteworthy, with a tendency to present them as technological solutions to complex societal problems without fully considering the underlying power dynamics and potential unintended consequences. The research emphasizes the need to actively shape the future of AVs, going beyond simply anticipating potential trajectories.

This paper contributes to the literature by offering a critical analysis of AV imaginaries in praxis. The findings show how different system actors use visualisations to shape perceptions of AV futures, highlighting the need for a more nuanced approach to interpreting these representations. Policymakers should actively shape AV futures, moving beyond simply anticipating future scenarios to actively determining desired outcomes. Future research should investigate the impact of sociotechnical imaginaries on decision-making processes and power dynamics in the development and deployment of AVs. It is also essential to consider the implications of the Covid-19 pandemic on these imaginaries.

The study is limited by the timeframe of the data (2015-2018), potentially missing significant shifts in landscape pressures and emerging technologies. The focus on visual analysis alone might not fully capture the complexities of sociotechnical imaginaries in practice. Future research should expand on the timeframe, investigate the impact of the Covid-19 pandemic on AV imaginaries, and explore the effects of these imaginaries beyond visual representations.