Spatial Computing in Next-Gen HCI for Smart Urban Planning

Abstract

Spatial computing—encompassing augmented reality (AR), virtual reality (VR), mixed reality (MR), and digital twin technologies—has evolved into a powerful Human–Computer Interaction (HCI) paradigm that offers immersive, context‑aware visualization and manipulation of urban environments. By overlaying geospatial data onto real‑world views or presenting fully synthetic 3D city models, spatial computing empowers planners, stakeholders, and community members to engage directly with design proposals in natural, intuitive ways. Yet despite considerable hype around “smart cities,” there remains a paucity of systematic evidence demonstrating how and to what extent spatial interfaces truly enhance planning efficiency, accuracy, and user satisfaction compared to conventional 2D GIS tools. In this study, we implemented a mid‑scale smart‑city model on a tabletop AR display and on the HoloLens 2 platform, then recruited thirty professional urban planners (mean experience = 6.2 years) to perform three prototypical tasks: (1) rezoning a multi‑parcel block, (2) siting transit stops based on population density, and (3) conducting line‑of‑sight analyses for proposed high‑rise developments. Using a within‑subjects counterbalanced design, we measured task completion time, spatial error rates, and perceived usability via the System Usability Scale (SUS). Statistical analysis via independent‑samples t‑tests revealed that spatial computing reduced mean completion time by 30% (12.4 ± 2.1 min vs. 17.8 ± 3.0 min; p < .001), halved error rates (0.6 ± 0.5 vs. 1.2 ± 0.7 errors/task; p = .002), and yielded SUS scores averaging 82.3 (excellent) versus 68.7 (OK) for 2D GIS (p < .001). We detail the system architecture, experimental methods, and quantitative outcomes, then discuss implications for municipal adoption, collaboration among diverse stakeholders, and the future integration of real‑time sensor feeds. Finally, we delineate scope and limitations—such as prototype maturity, dataset scale, and learning effects—and propose avenues for longitudinal, multi‑user, and large‑scale evaluations that can inform the next generation of smart‑city planning workflows.