Simulating human behavior through virtual agents is a key feature to improve the credibility of virtual environments (VE). For many use cases, such as daily activities data generation, having a good ratio between the agent's control and autonomy is required to impose specific activities while letting the agent be autonomous. This is why we propose a model allowing a user to configure the level of the agent's decisionmaking autonomy according to their requirements. Our model, based on a BDI architecture, combines control constraints given by the user, an internal model simulating human daily needs for autonomy, and a scheduling process to create an activity plan considering these two parts. Using a calendar, the activities that must be performed in the required time can be given by the user. In addition, the user can indicate whether interruptions can happen during the activity calendar to apply an effect induced by the internal model. The plan generated by our model can be executed in the VE by an animated agent in real-time. To show that our model manages well the ratio between control and autonomy, we use a 3D home environment to compare the results with the input parameters.