CoVR: Bridging Heterogeneous Physical Spaces for Collaborative VR via Personalized Scene Layouts

Collaborative tasks in Virtual Reality, such as jointly manipulating objects, require tight multi-user synchronization. However, users often operate in heterogeneous physical environments with varying room sizes, furniture layouts, and obstacles. Enforcing a single, globally consistent virtual scene across such diversity constrains users’ natural movements, underutilizes available physical space, and ultimately degrades collaboration efficiency. We present CoVR, a novel framework that enables seamless multi-user collaboration in VR across heterogeneous physical spaces by prioritizing interaction alignment over strict spatial alignment. Instead of requiring all users to share an identical virtual scene, CoVR provides each user with a personalized virtual scene layout while maintaining synchronized actions and consistent collaborative outcomes. Technically, we formulate a multi-user coupled optimization problem that jointly optimizes per-user layouts to maximize usable space and facilitate seamless task execution. A key challenge arises from bi-directional dependencies: one user’s optimized layout determines feasible trajectories, which in turn constrain other users’ optimizations, and vice versa. To tackle this, CoVR introduces an iterative joint optimization algorithm alternating between personalized scene refinement and cross-user synchronization until convergence. We evaluate CoVR across typical collaborative tasks. Results show that CoVR significantly outperforms conventional scene alignment methods, leading to notable improvements in spatial efficiency, task completion time, and user comfort. These findings underscore its potential for scalable, natural, and efficient multi-user VR collaboration.