IIC Journal of Innovation 10th Edition | Page 14

Intelligent Realities For Workers Using Augmented Reality, Virtual Reality and Beyond as tools that can import existing CAD drawings. In addition, AI can create 3D content. Booz Allen ® has demonstrated the use of generative adversarial networks to greatly reduce the time and expense of content creation for simulation. 16 computer vision is identifying parts of the bus, such as the tires, engine, etc. Mixing realities An additional type of SDK is focused on the mixing of digital and physical realities. To properly place a digital reality in to physical reality, the surfaces in the physical world and lighting need to be understood. ARCore™ from Apple ® and ARKit™ from Google solve these problems for iOS and Android, respectively. 17 World Knowledge with Computer Vision In Milgram’s paper cited earlier, the authors identified the dimension of Extent of World Knowledge, with the end points of an unmodelled world and a completely modelled world. For AR applications, an application needs to at least partially model the world, which the authors divided in to knowing what an object in the view is and where it is in the view. Intelligent Reality and AI The various technologies under the AI umbrella are quickly becoming just additional tools in the developer toolbox. This is certainly true in the XR space. For example, when the original Microsoft HoloLens released in 2016, it came with both voice recognition and computer vision available to application developers. 18 For enterprise reality applications, the identity of an object matters. For example, a fleet manager standing in front of a bus needs to know exactly what bus it is, not just that it is a bus. Object identity can be achieved by reading bar codes, QR codes, text or other uniquely identifying marks. But there is a lot more room for AI in the intelligent reality space than these baked-in features. The following sections look at several areas where AI can provide value to workers’ perceptions of their realities. Once an asset is identified, natural feature tracking and object recognition can be employed to recognize component parts. For example, the fleet manager can first glance at the license plate of the bus and then see the correct data-driven overlays for that bus as they move around because Overcoming XR UI Limitations Just as the transition from desktop to smart phone apps required new approaches, reality apps introduce their own UI 16 N. Mehta et al., “The Role of AI in a VR World,” GPU Technology Conference, October 2018. Available: http://on- demand.gputechconf.com/gtcdc/2018/pdf/dc8209-the-role-of-ai-in-a-vr-world-presented%20by-booz-allen-hamilton.pdf 17 M. Giles, “AR still doesn’t have a killer app, but Google’s ARCore is here to help,” MIT Technology Review, Feb 2018. Available: https://www.technologyreview.com/s/610336/ar-still-doesnt-have-a-killer-app-but-googles-arcore-is-here-to-help/ 18 D. Bohn and T. Warren, “Up close with the HoloLens, Microsoft’s most intriguing product in years,” The Verge, January, 2015 [online]. Available: https://www.theverge.com/2015/1/21/7868251/microsoft-hololens-hologram-hands-on-experience IIC Journal of Innovation - 10 -