Virtual reality based interior design recommender system using rule based algorithms

One of the problems of interior design decision making is that homeowners cannot picture furniture in their real space before they decide on buying and arranging it. Besides that, current design tools are mostly 2D-based and do not offer an immersive spatial experience, which restricts users to making decisions based on 2D drawing. In this project, a Virtual Reality bedroom layout recommender system was designed and implemented as a standalone VR system for the Meta Quest 2 headset. The system was built with Unity2022.3, XR Interaction Toolkit and a Rule-Based Inference Engine developed in C# and based on nine rules, related to the use of spatial and thematic placement of furniture, including geometric fit, wall clearance, relation proximity, and design theme consistency. A total of three themes of interior design were supported: Minimalist, Modern, and Scandinavian, and these themes were applied to three different room layouts: a square room, a rectangular room, and an L-shaped room. A virtual world interface was used to present a furniture catalogue, allowing the users to navigate in the first-person perspective, while a system of scores prioritised the recommendations for furniture placements presented to users automatically. The six basic system functions were verified working by functional testing which include furniture selection, rule-based placement, locomotion and rendering performance at a stable 72 frames per second. A Rule Compliance Rate of 96.67% were achieved in accuracy testing, and the few non-compliant placements occurred because certain closet configurations required rotation, reducing the available door clearance. This limitation has been identified for future improvement. The overall mean satisfaction score obtained from the User Acceptance Testing of 48 users was 4.07 of 5.00, which suggests that the users accepted the product on usability, visual quality, and perceived usefulness. The system shows that rule-based inference embedded in an immersive Virtual Reality environment can be used to effectively assist furniture placement decisions and suggests further work to expand the scope of the furniture themes, add adaptive rule weighting, and eliminate any remaining constraints of the clearance of closets.