Opinion examines barriers to XR adoption in engineering

AI-generated summary

XR adoption in engineering has been constrained less by headsets themselves than by workflows that simplify CAD models to fit their limited computing power, argues Florian Haspinger, cofounder and managing director of Hololight, in an opinion article for Engineering.com. Immersive reviews can bridge the gap between viewing designs on a flat screen and building expensive physical prototypes. However, complex engineering models often exceed a headset’s rendering capacity. Reducing their complexity can remove precisely the details reviewers need to understand a design in space. Haspinger’s central argument is that rendering should move off the headset rather than requiring engineers to compromise the model. The article traces immersive engineering review back to the CAVE, developed at the University of Illinois Chicago in 1992. This room-sized installation projected imagery onto its walls, floor and ceiling. Viewers wore LCD shutter glasses with a position sensor, allowing the projections to adjust to their perspective as they moved. Rendering took place on the system driving the projectors, not on the glasses. Major automotive and aerospace companies led adoption, but high costs and substantial space requirements limited accessibility. Because the computing infrastructure was tied to a particular room, reviews also remained location-dependent and could require travel. Portable, lower-cost XR headsets addressed some of those access problems without eliminating model preparation. Both earlier systems and headsets required engineers to perform tasks such as tessellating geometry, reducing polygon counts and removing hidden parts. Haspinger emphasizes that the resulting review asset is a derivative rather than the model of record. Preparing it takes time, while the original CAD design continues to evolve. Consequently, the version available for review can lag behind current engineering work, and transferring annotations and identified problems back into the original data adds further versions to manage. This preparation also creates uncertainty about review findings. When reviewers identify an apparent issue, they may not know whether it exists in the original design or was introduced by simplification. Haspinger argues that this weakens confidence in engineering decisions and helps explain why some teams abandoned XR in favor of conventional screen-based reviews, where they could at least display the complete model correctly. In his account, successive generations of immersive technology changed the hardware but retained a recurring compromise between accessibility, model fidelity and review reliability. XR streaming offers an alternative architecture. The application containing the model continues running on its existing computing system, which renders the design at full complexity. Pixels are streamed to the headset while the original geometry remains where it is. Haspinger argues that this allows teams to review an assembly as soon as it is modeled, including materials and lighting, without preparing a simplified asset or reconciling a separate review version afterward. The claimed benefits include same-day reviews of revisions, faster iteration and reduced reliance on costly physical prototypes. Streaming also supports remote participation, potentially reducing travel across teams and locations. Spatial presentation can help customers and other stakeholders unfamiliar with CAD understand a design more quickly than a two-dimensional display, improving communication and helping expose issues before they become late-stage changes. However, the approach shifts technical requirements rather than removing them: organizations must establish what the connection between the rendering machine and headset needs before scheduling a review. Rendering hardware must also be adequate for the application, model complexity and number of concurrent sessions. Haspinger acknowledges that XR is unnecessary for some workflows and review stages. His conclusion is that streaming can make it an on-demand review tool, organized around a design question rather than hardware limitations. For engineering data management, the key argument concerns maintaining fidelity to the actual design and avoiding derivative-version reconciliation. These benefits are presented as the author’s position, not as results from comparative testing or quantified deployments. His company, Hololight, provides industrial XR infrastructure that separates software from headset hardware and streams pixels rather than model data.