We have successfully designed and implemented a multi-player, multi-modal virtual reality (VR) co-simulator that integrates both pedestrians and cyclists within the same VR environment. This co-simulator is designed to support multi-player, multi-modal VR experiments aimed at comprehensive data collection in transportation research. It features advanced functionalities including body tracking for pedestrians and cyclists, high-quality digital human representations, and comprehensive data collection technologies such as eye gazing data. A VR experiment was conducted to evaluate the effectiveness of the VR co-simulator and investigate the interactions between vulnerable road users (VRUs) and automated vehicles (AVs) in shared spaces. The assessment demonstrated the co-simulator's capabilities and feasibility in terms of simulator sickness, presence, realism, and usability. The experiment also confirmed the impact of VRU combinations and initial relative positions on the interactions between VRUs and AVs in shared space.