Human-robot collaboration (HRC) in shared workspaces has become increasingly common in real-world applications and has drawn considerable research attention. However, most existing studies on human-in-the-loop (HITL) collaboration with robots are conducted either in simplified game environments or on physical platforms, which limits their realism and scalability. To address this, we propose an embodied framework called HumanTHOR, which allows humans to interact with robots in a simulation environment using VR devices, facilitating HITL collaboration in a shared workspace. To validate the system, we develop a benchmark of everyday tasks and conduct a preliminary user study using two baseline algorithms. The results demonstrate that the robot can effectively assist humans in collaboration, highlighting the significance of HRC. A comparison of different baseline levels confirms that our system can assess robot capabilities and serve as a benchmark for evaluating various robot algorithms. Additionally, the experimental results suggest that there is still substantial potential in this area, and our system provides a solid foundation for future HRC research in shared workspaces.