Navigation systems and methods for tracking physical objects near a target site during a surgical procedure. The navigation system comprises a robotic device and an instrument attached to the robotic device. A vision device is attached to the robotic device or the instrument and generates vision data sets. The vision data sets are captured from multiple perspectives of the physical object. A computing system associates a virtual object with the physical object based on one or more features of the physical object identifiable in the vision data sets. The virtual object at least partially defines a virtual boundary for the instrument.